Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
quantum correlations , which lie in the foundations of quantum theory , have been of renewed interest during the last two decades as the field of quantum information science emerged and matured . quantum entanglement , as a kind of quantum correlations , has been playing central roles in quantum information and computation @xcite . the negativity is one of the best known and most popular tools of quantifying bipartite quantum entanglement . it can be computed easily for arbitrary states of a composite system . therefore , it has received various studies @xcite . however , quantum entanglement does not account for all of the non - classical properties of quantum phenomenons . therefore , numerous quantifiers of quantum correlations have been further proposed to reveal the non - classical correlations that can not be fully captured by quantum entanglement @xcite . among the various measures , it has been shown that quantum discord provides a larger region of quantum states with non - classical correlations , and a non - zero discord quantum state without entanglement may be responsible for the efficiency of a quantum computer . therefore , much attention has been paid to study the quantum discord in various quantum systems @xcite . however , due to the complicated optimization , it is usually complicated to calculate quantum discord analytically . even for qubit - qubit quantum systems , the analytical formula of quantum discord can only be obtained for a few cases , and a general method is still lacking . as a variant of quantum discord , geometric discord ( gd ) was introduced by dakic and coworkers @xcite . due to its simplicity in calculation , it has attracted a lot of research interests @xcite . it is well known that quantum systems inevitably interact with the environment when quantum information is processed in the real - world , which brings about the loss of initially presented quantum properties @xcite . this phenomenon can be characterized by decoherence , which can be illustrated by the example of faithfully transmitting an unknown quantum state through a noisy quantum channel . during the transmission , the carrier of the information interacts with the channel and gets correlated with other degrees of freedom . this gives rise to the phenomenon of so - called decoherence on the subspace of the information carrier . recent studies on open quantum systems highlight the existence of two different classes of dynamical behaviours known as markovian and non - markovian regimes @xcite . in the quantum domain , under certain assumptions , markovian dynamics lead to a master equation in the lindblad form . however , in certain scenarios such approximations are not justified and one needs to go beyond perturbation theory . it is clear that , due to the general complexity of the problem to be studied , exact solutions exist only for simple open quantum systems models such as the well - known jaynes - cummings model , quantum brownian motion model , and certain pure dephasing models @xcite . for the deeper understanding of quantum phenomenons , it is desirable to investigate the behaviours of the quantum properties under the action of decoherence . the previous work showed that quantum discord and entanglement behave differently under the effect of the environment @xcite . in particular , for certain quantum states , the phenomenon of entanglement sudden death does not occur for quantum discord . there exists some studies on the quantum correlations of qubit - qubit and qubit - qutrit systems under the effect of local quantum noises in the markovian environment @xcite . the importance of quantum states with higher dimensions is gradually recognized in recent years @xcite . therefore , a deep understanding of the quantum correlations dynamics of the high - dimensional quantum systems beyond qubits is very practical . song et al . @xcite have studied the distillability in qutrit - qutrit systems under local dephasing channels , and found that the distillability will be decrease after a finite time . m. ali studied the distillability sudden death in qutrit - qutrit systems under global and multilocal dephasing @xcite , and amplitude damping channels @xcite . he has found that the quantum entanglement will decrease after a finite time under dicoherence . guo et al . @xcite studied the dynamics of quantum correlations of qubit - qutrit systems under various decoherent channels . they have shown that the decoherent channels bring with different influences for the dynamics of quantum correlations measured by negativity , quantum discord and gd . the influences depend on the initial state parameters and the properties of the decoherent channels . in this work , we will study the quantum correlations in qutrit - qutrit systems under identical and non - identical local noise channels . we consider four kinds of local noise channels on each qutrit , including dephasing , trit - flip , trit - phase - flip , and depolarising channels . the remainder of this paper is organized as follows : in sec . [ sec:2 ] , we will introduce the negativity and the gd as our measures of quantum correlations . in sec . [ sec:3 ] , the evolution of the quantum correlations for a qutrit - qutrit system under the identical and non - identical local noise channels are discussed . finally , we conclude in sec . [ sec:4 ] . in this section , we will describe the gd and the negativity in a brief manner . the negativity is a useful entanglement measure for arbitrary bipartite state quantum systems @xcite , especially for free entangled states . for a given density matrix @xmath0 , its negativity will be @xmath1 where @xmath2 is the partial transpose of @xmath0 with respect to subsystem @xmath3 and @xmath4 represents the trace norm . if @xmath5 , then the bipartite state is free entangled . [ [ gd ] ] gd ~~ geometric measure of quantum discord in a bipartite state can be described as the distance of the state from the closest zero - discord state @xcite @xmath6 where @xmath7 is the hilbert - schmidt norm and @xmath8 is the set of all zero discord states . for a general bipartite state of @xmath9 dimensions , the density matrix can be written as @xmath10,\ ] ] where @xmath11 ( @xmath12 ) and @xmath13 ( @xmath14 ) are the generators of @xmath15 and @xmath16 groups ( special unitary groups of dimensions @xmath17 and @xmath18 , respectively ) , satisfying the relation @xmath19 . the lower bound @xcite of the gd has been derived as @xmath20 where @xmath21 with bloch vector @xmath22 with elements @xmath23 , matrix @xmath24 is the correlation matrix with elements @xmath25 and @xmath26 are the eigenvalues of the matrix @xmath27 arranged in non - increasing order . we will use the lower bound as the estimation of the gd in the following . let us consider a qutrit - qutrit system with each particle interacts independently with the local noise channels . in this case , the evolution of a quantum state can be described by the lindblad equation , which can be written in terms of the kraus operators as @xmath28 where @xmath29 and @xmath30 are the kraus operators characterizing the local noise channels on @xmath3 and @xmath31 , respectively . they satisfy @xmath32 and @xmath33 @xcite . to study the dynamics of quantum correlations for a qutrit - qutrit system , we choose the bell state @xmath34 as the initial quantum state . in the following , we will study the dynamics of the quantum correlations of @xmath35 under the effect of four different noise channels . physically , dephasing corresponds to any process of losing coherence without the exchange of energy . under the action of dephasing noises , the off - diagonal elements of the density matrix decay exponentially against time . for two local dephasing noises on qutrits @xmath3 and @xmath31 , the kraus operators are given by @xmath36 where @xmath37 @xmath38 or @xmath31 , @xmath39 and @xmath40 with @xmath41 and @xmath42 denoting the decay rates of qutrits @xmath3 and @xmath31 , respectively . for trit - flip channels of qutrits @xmath3 and @xmath31 , the kraus operators are given by @xmath43 where @xmath44 the kraus operators are @xmath45 where @xmath46 the kraus operators are @xmath47 where @xmath48 based on the numerical simulations , we give our results in figs . [ fig1 ] , [ fig2 ] , [ fig3 ] , and [ fig4 ] for each two identical dephasing , trit - flip , trit - phase - filp , and depolarizing channels , respectively . one can observe that under the effect of two identical noise channels , as @xmath49 or @xmath50 increases , the negativity and the gd decrease rapidly ( see @xmath51 and @xmath52 in the figures ) , which indicates that quantum correlation is strongly affected by large decay rates @xcite . by observing figs . [ fig1 ] and [ fig2 ] , one can see that the gd is more robust than the negativity ageist two identical dephasing and trit - flip noise channels . however , as shown in figs . [ fig3 ] and [ fig4 ] , for two identical tirt - phase - flip and depolarizing channels , the negativity becomes more robust than the gd . to study the effect of the decay rates of the channels , we draw the negativity and the gd versus the decay rates @xmath49 or @xmath50 in three - dimension figures with a fixed time @xmath53 ( see @xmath51 and @xmath52 in the figures ) . it can been seen that in the case where @xmath49 and @xmath50 are large , the negativity and the gd are influenced strongly during the decoherence . in this section , we discuss the negativity and the gd under non - identical local noise channels , namely , the noise channels are of different types and decay rates . the kraus operators have been given in sec . [ idt ] . the results are drawn in figs . [ fig5 ] , [ fig6 ] , [ fig7 ] , [ fig8 ] , [ fig9 ] , and [ fig10 ] for dephasing - trit flip channels , dephasing - trit phase flip channels , dephasing - depolarising channels , trit flip - trit phase flip channels , trit flip - depolarising channels , and trit phase flip - depolarising channels respectively . it can be seen that under the effect of two different local noise channels , the negativity and the gd behave similar to the case where the channels are identical . one can observe that in all cases of non - identical local noise channels the gd is more robust than the negativity . same in the identical channels , as the time elapsed the state losses its quantum entanglement . to study just the effect of rate of the channels we draw the negativity and the gd versus the rates of two channels in three dimensions at a fixed time @xmath53 ( see @xmath54 and @xmath55 in the figures ) . as the rates of the channels are large , the negativity and the gd decease rapidly . to summarize , in this work , we have studied the quantum correlations of a qutrit - qutrit quantum system under identical and non - identical local noise channels by using the negativity and the gd . we have considered four kinds of local noise channels on each qutrit , including dephasing , trit - flip , trit - phase - flip , and depolarising channels . meanwhile , we have further investigated the effect of various combinations local noise channels on the quantum correlation . the results show that the quantum correlations decrease monotonically with time under decoherence . meanwhile , when the time is fixed , negativity and geometric discord decrease rapidly when the decay rates are large . we expect our work may find further theoretical and experimental applications . bostrom , k. , felbinger , t. : phys . lett . * 89 * , 187902 ( 2002 ) . yu , g. : int . . phys . * 51 * , 2954 ( 2012 ) . deng , f. g. , li , x. h. , at all . a * 72 * , 359 ( 2006 ) . yang , y. g. , wen , q. y. : sci . china ser . g 5 * 50 * , 558 ( 2007 ) . guhne , a. , hyllus , p. : int . j. theor 42 * , 1001 ( 2003 ) . nakano , t. , piani , m. , adesso , g. : phys . a * 72 * , 012117 ( 2013 ) eltschka , c. , siewert , j. : phys . lett . * 111 * , 100503 ( 2013 ) . datta , a. : phys . a * 81 * , 052312 ( 2010 ) . ferrie , c. , morris , r. , emerson , j. : phys . a * 82 * , 044103 ( 2010 ) . dajka , j. , mierzejewski , m. , uczka , j. , blattmann , r. , hanggi , p. : j. phys . a : math . theor . * 45 * , 485306 ( 2012 ) . nielson , m.a . , chuang , i.l . : cambridge university press , cambridge ( 2000 ) . rudolph , o. : phys . a * 67 * , 032312 ( 2003 ) . moradi , s. : quant . inf . comp . * 11 * , 957 ( 2011 ) . , song , h.s . a * 72 * , 022333 ( 2005 ) . vidal , g. , werner , r.f . a * 65 * , 032314 - 032317 ( 2002 ) . horodecki , r. , horodecki , p. , horodecki , m. , horodecki , k. : rev . phys . * 81 * , 865 ( 2009 ) ollivier , h , . zurek , w.h . lett . * 88 * , 017901 ( 2001 ) . brodutch , a , . terno , d.r . a * 83 * , 010301(r ) ( 2011 ) . datta , a. , shaji , a. , caves , c.m . . lett . * 100 * , 050502 ( 2008 ) . lanyon , b.p . , barbieri , m. , almeida , m.p . , white , a.g . : * 101 * , 200501 ( 2008 ) . okrasaa , m.,walczak , z. : euro . . lett . * 96 * , 60003 ( 2011 ) . zhang . y. , zhao . h. : quantum inf . process . ( published online ) . he , j. , tao liu . , w. : int . j. theor . * 52 * , 3381 ( 2013 ) . liu , x. , ma , j. , xi , z.,wang , x. : phys . a * 83 * , 012327 ( 2011 ) . chen , q. , zhang , c. , yu , s. , yi , x.x . , oh , c.h . : phys . rev . a * 84 * , 042313 ( 2011 ) . ali , m. : j. phys . a : math . theor . * 43 * , 495303 ( 2010 ) . dakic , b. , vedral , v. , brukner , c. : phys . rev . lett . * 105 * , 190502 ( 2010 ) . rana , s. , parashar , p. : phys . a * 85 * , 024102 ( 2012 ) . vidal , g. , werner , r. f. : phys . a * 65 * , 032314 ( 2002 ) . doustimotlagh , n. , wang , s. , you , c. , long , g. l. : epl , * 106 * , 60003 ( 2014 ) . wang , j. , jing , j. : phys . a * 82 * , 032324 ( 2010 ) . wang , j. , jing , j. : ann . phys . * 327 * , 283 ( 2012 ) . hu , m.l . , fan , h. : ann . phys . * 327 * , 851 ( 2012 ) . miao , c. , yang , m. , cao , z , l. : int . j. theor . phys . * 52 * , 1780 ( 2013 ) . ramzan , m. , khan , m.k . inf . process . * 11 * , 443 ( 2012 ) . ren , b.c . , wei , h.r . , deng , f.g . : quantum inf . process . * 13 * , 1175 ( 2014 ) . guo , j.l . , li . h. , long . g. l. : quantum inf . process . * 12 * , 3421 ( 2013 ) . chou , c.h.,yu , t. , hu , b.l . e * 77 * , 011112 ( 2008 ) . an , j.h . , zhang , w.m . a * 76 * , 042127 ( 2007 ) . an , j.h . , feng , m. , zhang , w.m . : quantum inf . * 9 * , 0317 ( 2009 ) . c. , bylicka.b . , chruscinski . d. , maniscalco . s. : arxiv:1402.4975v3 . li , j.g . , zou , j. , shao , b. : phys . a * 82 * , 042318 ( 2010 ) . maziero , j. , cleri , l.c . , serra , r.m . , vedral , v. : phys . rev . a * 80 * , 044102 ( 2009 ) . fanchini , f.f.,werlang , t. , brasil , c.a . , arruda , l.g.e . , caldeira , a.o . a * 81 * , 052107 ( 2010 ) . d. : arxiv:1403.2446v1 . wang , s.,yao , l. , long , g.l.:phys . a * 87 * , 062305 ( 2013 ) . song , w. , chen , l. , zhu , s , l. : phys . a * 80 * , 012331 ( 2009 ) . mazhar , a. : phys . rev . a * 81 * , 042303 ( 2010 ) . mazhar , a. : j. phys . b : at . mol . phys * 43 * , 045504 ( 2010 ) . hassan , a.s.m . , lari , b. , joag , p.s . a * 85 * , 024302 ( 2012 ) . xu , j. : j. phys . * 45 * , 405304 ( 2012 ) . hassan , a.s.m . , joag , p.s . : j. phys . a : math . theor . * 45 * , 345301 ( 2012 ) .	due to decoherence , realistic quantum systems inevitably interact with the environment when quantum information is processed , which causes the loss of quantum properties . as a fundamental issue of quantum properties , quantum correlations have attracted a lot of interests in recent years . because of the importance of high dimensional systems in quantum information , in this work , we study the quantum correlations affected by the markovian environment by considering the quantum correlations of qutrit - qutrit quantum systems measured by the negativity and the geometric discord . the local noise channels covered in this work includes dephasing , trit - flip , trit - phase - flip , and depolarising channels . we have also investigated the cases where the local decoherence channels of two sides are identical and non - identical .
the standard model is flawed by the large number of free parameters , for which there is at present no explanation . there is no prediction of the family replication pattern , nor of the number of families . all the families are really treated on the same footing . most of the standard model free parameters reside in `` flavour space '' - with six quark masses , six lepton masses , four quark mixing angles and ditto for the leptonic sector , as well as the strong cp - violating parameter @xmath0 . the structure of flavour space is determined by the fermion mass matrices , i.e. by the form that the mass matrices take in the `` weak interaction basis '' where mixed fermion states interact weakly , in contrast to the `` mass bases '' , where the mass matrices are diagonal . one may wonder how one may ascribe such importance to the different bases in flavour space , considering that the information content of a matrix is contained in its matrix invariants , which in the case of a @xmath1 matrix @xmath2 are the @xmath3 sums and products of the eigenvalues @xmath4 , such as @xmath5 , @xmath6 , @xmath7 these expressions are invariant under permutations of the eigenvalues , which in the context of mass matrices means that they are flavour symmetric , and obviously independent of any choice of flavour space basis . even if the information content of a matrix is contained in its invariants , the form of a matrix may also carry information , albeit of another type . the idea - the hope - is that the form that the mass matrices have in the weak interaction basis can give some hint about the origin of the unruly masses . there is a certain circularity to this reasoning ; to make a mass matrix ansatz is in fact to define what we take as the weak interaction basis in flavour space . we denote the quark mass matrices of the up- and down - sectors in the weak interaction basis by @xmath2 and @xmath8 , respectively . we go from the weak interaction basis to the mass bases by rotating the matrices by the unitary matrices @xmath9 and @xmath10 , @xmath11 @xmath12 the lodestar in the hunt for the right mass matrices is the family hierarchy , with two lighter particles in the first and second family , and a much heavier particle in the third family . this hierarchy is present in all the charged sectors , with fermions in different families exhibiting very different mass values , ranging from the electron mass to the about @xmath13 times larger top mass . it is still an open question whether the neutrino masses also follow this pattern @xcite . in the democratic approach @xcite , @xcite , @xcite the family hierarchy is taken very seriously . it is assumed that in the weak basis the fermion mass matrices have a form close to the @xmath14 symmetric `` democratic '' matrix @xmath15 with the eigenvalues @xmath16 , reflecting the family hierarchy . the underlying philosophy is that in the standard model , where the fermions get their masses from the yukawa couplings by the higgs mechanism , there is no reason why there should be a different yukawa coupling for each fermion . the couplings to the gauge bosons of the strong , weak and electromagnetic interactions are identical for all the fermions in a given charge sector , it thus seems like a natural assumption that they should also have identical yukawa couplings . the difference is that the weak interactions take place in a specific flavour space basis , while the other interactions are flavour independent . the democratic assumption is thus that the fermion fields of the same charge initially have the same yukawa couplings . with three families , the quark mass matrices in the weak interaction basis then have the ( zeroth order ) form @xmath17 where @xmath18 and @xmath19 have dimension mass . the corresponding mass spectra @xmath20 reflect the family hierarchy with two light families and a third much heavier family , a mass hierarchy that can be interpreted as the representation @xmath21 of @xmath22 . in order to obtain realistic mass spectra with non - zero masses , the @xmath14 symmetry must obviously be broken , and the different democratic matrix anstze correspond to different schemes for breaking the democratic symmetry . we can apply the democratic approach to the lepton sector as well , postulating democratic ( zeroth order ) mass matrices for the charged leptons and the neutrinos , whether they are fermi - dirac or majorana states , @xmath23 relative to the quark ratio @xmath24 , the leptonic ratio @xmath25 is so extremely small that it seems unnatural . one way out is to simply assume that @xmath26 vanishes , meaning that the neutrinos get no mass contribution in the democratic limit @xcite . according to the democratic philosophy , then there would be no reason for a hierarchical pattern la the one observed in the charged sectors ; the neutrino masses could even be of the same order of magnitude . data are indeed compatible with a much weaker hierarchical structure for the neutrino masses than the hierarchy displayed by the charged fermion masses . unlike the situation for the quark mixing angles , in lepton flavour mixing there are two quite large mixing angles and a third much smaller mixing angle , these large mixing angles can be interpreted as indicating weak hierachy of the neutrino mass spectrum . the neutrino mass spectrum hierarchy could even be inverted ; if the solar neutrino doublet @xmath27 has a mean mass larger than the remaining atmospheric neutrino @xmath28 , the hierarchy is called `` inverted '' , otherwise it is called `` normal '' . supposing that the neutrino masses do not emerge from a democratic scheme , a ( relatively ) flat neutrino mass spectrum could be taken as a support for the idea that the masses in the charged sectors emerge from a democratic scheme in the case that both the up- and down - sector mass matrices have a purely democratic texture , the quark mixing matrix is @xmath29 , where @xmath30 is the unitary matrix that diagonalizes the democratic matrix ( [ nambu ] ) . we use this to define the democratic basis , meaning the flavour space basis where the mass matrices are diagonalized by ( [ dem ] ) and the mass lagrangian is symmetric under permutations of the fermion fields @xmath31 of a given charge sector . in the democratic basis the mass lagrangian @xmath32 is symmetric under permutations of the fermion fields @xmath31 , while in the mass basis with @xmath33 the mass lagrangian has the form @xmath34 which is clearly not invariant under permutations of ( @xmath35,@xmath36,@xmath36 ) . we can perform a shift of the democratic matrix , by just adding a unit matrix @xmath37 , @xmath38 , @xmath39 corresponding to the mass spectrum @xmath40 . the matrix @xmath41 has a democratic texture , both because it is diagonalized by @xmath42 , and because the mass lagrangian is invariant under permutations of the quark fields , @xmath43 if @xmath41 and @xmath44 both have a texture like ( [ pert ] ) , there is no cp - violation . this is independent of how many families there are , because of the degeneracy of the mass values . cp - violation only occurs once there are three or more non - degenerate families , because only then the phases can no longer be defined away . we can repeat the democratic scheme with a number @xmath45 of families , where the fermion mass matrices again are proportional to the @xmath46 symmetric democratic matrix which is diagonalized by a unitary matrix analogous to @xmath42 in ( [ dem ] ) . to the @xmath47-dimensional democratic matrix term , we can again add a @xmath47-dimensional diagonal matrix @xmath48 , and get a @xmath47-dimensional mass spectrum with @xmath45 massive states , and @xmath49 degenerate masses . the mass matrix still has a democratic texture , and there is still no cp - violation . in order to obtain non - degenerate , non - vanishing masses for the physical flavours @xmath50 , the permutation symmetry of the fermion fields @xmath31 in the democratic basis must be broken . the proposal here is to derive mass matrices with a nearly democratic texture , not by explicitly perturbing the assumed initial democratic form ( [ nambu ] ) , but instead by perturbing the matrix @xmath42 which diagonalises the democratic mass matrix . this is done by deriving the unitary rotation matrices @xmath9 , @xmath10 for the up- and down- sectors , from a specific parameterisation of the weak mixing matrix @xmath51 . the idea is to embed the assumption of democratic symmetry into the standard model mixing matrix , by expressing the mixing matrix as a product @xmath52 since both the mixing matrix and its factors , according to the standard parameterisation @xcite , are so close to the unit matrix , the rotation matrices @xmath9 , @xmath10 are effectively perturbations of the democratic diagonalising matrix ( [ dem ] ) . in this way , the weak interaction basis remains close to the democratic basis . the cabbibo - kobayashi - maskawa ( ckm ) mixing matrix @xcite can of course be parametrized - and factorized - in many different ways , and different factorizations correspond to different rotation matrices @xmath9 and @xmath10 . the most obvious and `` symmetric '' factorization of the ckm mixing matrix is , following the standard parametrization @xcite with three euler angles @xmath53 , @xmath54 , @xmath55 , @xmath56 with the diagonalizing rotation matrices for the up- and down - sectors @xmath57 and @xmath58 respectively , where @xmath53 , @xmath54 , @xmath59 and @xmath60 correspond to the parameters in the standard parametrization in such a way that @xmath61 , @xmath62 rad , and @xmath63 , while @xmath64 and @xmath65 . from the rotation matrices @xmath9 and @xmath10 we then obtain the mass matrices @xmath66 and @xmath67 , such that @xmath68 where @xmath69 is the trace @xmath70 , and with @xmath71 , @xmath72 , @xmath73 , @xmath74 @xmath75 @xmath76 @xmath77\hspace{1mm}+\frac{f}{2}\hspace{1mm}(d - c)\hspace{1mm}(e^{i\gamma}-2\hspace{1mm}e^{-i\gamma})-2t+6\hspace{1mm}m_u$ ] @xmath78 similarly for the down - sector , @xmath79 with the parameters @xmath80 , @xmath81 , @xmath82 , @xmath83 , and @xmath84 @xmath85 @xmath86 @xmath87+\frac{f'}{2}(j+g)(2e^{i\gamma}-e^{-i\gamma})-2t'+6m_b$ ] @xmath88 in order to evaluate to what degree these rather opaque matrices are democratic , we calculate numerical matrix elements by inserting numerical mass values . for the up - sector we get the ( nearly democratic ) matrix texture @xmath89\ ] ] where the `` small '' matrix @xmath90 with @xmath91 , is what breaks the democratic symmetry , supplying the two lighter families with non - zero masses . with mass values calculated at @xmath92 ( jamin 2014 ) @xcite , @xmath93 we get @xmath94 , @xmath95,@xmath96 mev @xmath97 , and @xmath98 , @xmath99,@xmath100 , @xmath101,@xmath102 . for the down - sector , with @xmath103 we get another democratic texture , @xmath104 where @xmath105 , @xmath106 , @xmath107 , @xmath108 , and @xmath109 @xmath110 @xmath111 . just like in the up - sector mass matrix , the matrix elements in @xmath8 display a nearly democratic texture . for both the up - sector and the down - sector the mass matrices are thus approximately democratic . in the mass matrix literature there is an emphasis on `` calculability '' . the ideal is to obtain mass matrices that have a manageable form , but there is nothing that forces nature to serve us such user - friendly formalism . it is however tempting to speculate that there are relations between the elements that could make the democratic matrices more calculable , and in the search for matrices that are reasonably transparent and calculable , we look at a more radical factorization of the mixing matrix , viz . @xmath112 and @xmath113 where , as before , @xmath62 rad , and @xmath114 , while @xmath64 , and @xmath65 . these rotation matrices are still `` perturbations '' of the democratic diagonalizing matrix ( [ dem ] ) , and the up - sector mass matrix has a texture similar to ( [ 1 ] ) , @xmath115 where @xmath69 is the trace , @xmath70 , and @xmath116 , @xmath117 , @xmath118 @xmath119 , @xmath120 with @xmath121 , @xmath122 . this matrix can be reformulated in a form similar to ( [ cu ] ) , @xmath123\ ] ] where @xmath124 , @xmath125 $ ] , @xmath126 , and @xmath127 with @xmath128,@xmath129 , @xmath130 , @xmath131 , @xmath132 inserting the mass values ( [ jam1 ] ) gives @xmath133,@xmath134 , @xmath135 , and @xmath136 , @xmath137 , @xmath138 for the down - sector , with @xmath139 the mass matrix @xmath140 reads @xmath141 where @xmath142 @xmath143 @xmath144 , @xmath145 . inserting the mass values ( [ jam2 ] ) we moreover get the numerical values @xmath146,@xmath147,@xmath148,@xmath149 . by including the democratic rotation matrix in the parametrization of the weak mixing matrix , we obtain mass matrices with specific democratic textures . in this way we make contact between the democratic hypothesis and the experimentally derived parameters of the ckm mixing matrix , avoiding the introduction of additional concepts . h. fusaoka , y. koide , `` updated estimate of running quark masses '' , hep - ph/9712201v2 , dec . 1997 ; r. barbieri , g. dvali , a. strumia , z. berezhiani , l. hall , `` flavour in supersymmetric grand unification : a democratic approach , '' hep - ph/9405428v3 .	within the framework of quark mass matrices with a democratic texture , the unitary rotation matrices that diagonalize the quark matrices are obtained by a specific parametrization of the cabibbo - kobayashi - maskawa mixing matrix . different forms of democratic quark mass matrices are derived from slightly different parametrizations .
in this paper , we consider the relation between jamming ( a kinetic process ) and the laws of static mechanical equilibrium of particulate media . we describe first a simple model of jamming in a colloid , sheared between parallel plates . we assume that force chains ( linear arrays of rigid particles in contact ) develop along the major stress compression axis and span the sample . the resulting jammed state can support a shear stress indefinitely , but does this by virtue of a self - organized internal structure ( the force chain array ) which has developed in direct response to the applied load itself . if a different load is now applied ( e.g. if the material is sheared in some different direction ) , the force chains can not sustain the load but must flow and rejam in a different configuration . the model thus provides a simple example of a fragile " material one which can not support certain types of infinitesimal stress increment without plastic reorganisation . the continuum mechanics of fragile materials is very different from conventional elastic or elastoplastic theory . we propose that fragility may be generic in jammed systems , and argue further that a pile of cohesionless poured sand can be viewed as jammed in the required sense : its mechanical integrity results purely from the applied load ( gravity ) . some simple models that we have recently developed for poured sand @xcite , such as the fixed principal axes ( fpa ) model , indeed display fragile behaviour ; in fact they correspond to assuming a particular simplified geometry for the granular skeleton of force chains . fragile models require that the behaviour of a pile of cohesionless grains ( fragile ) is quite different from a hypothetical pile where each grain is firmly glued to its neighbours on first coming to rest ( elastic ) . this is plausible because for random packings of glued grains , a finite fraction of the interparticle forces will be under tension ( forbidden in the cohesionless case ) and will remain so until the load is large enough to cause appreciable elastic deformation of individual grains . for simple fragile models such as fpa , the same criterion defines a crossover between fragile behaviour and a form of ( anisotropic ) elastoplasticity . a fuller discussion of these ideas is given in @xcite , where further references may be found . consider a concentrated colloidal suspension of hard particles , confined between parallel plates at fixed separation , to which a shear stress is applied . above a certain threshold of stress , this system exhibits enters a regime of strong shear thickening ; see , e.g. @xcite . the effect can be observed in the kitchen , by stirring a concentrated suspension of corn - starch with a spoon . in fact , computer simulations suggest that , at least under certain idealized conditions , the material will jam completely and cease to flow , no matter how long the stress is maintained @xcite . in these simulations , jamming apparently occurs because the particles form force chains " @xcite along the compressional direction ( fig . [ fig : jammed ] ( a ) ) . even for spherical particles the lubrication films can not prevent direct interparticle contacts ; once these arise , an array or network of force chains can indeed support the shear stress indefinitely . ( we ignore brownian motion , here and below , as do the simulations ; this could cause the jammed state to have finite lifetime . ) to model the jammed state , we start from a simple idealization of a force chain : a linear string of at least three rigid particles in point contact . crucially , this chain can only support loads _ along its own axis _ ( fig.[fig : paths ] ( a ) ) : successive contacts must be collinear , with the forces along the line of contacts , to prevent torques on particles within the chain @xcite . note that neither friction at the contacts , nor particle aspherity , can change this longitudinal force " rule . as a minimal model of the jammed colloid , we take an assembly of such force chains , characterized by a unique director @xmath0 , in a sea of spectator " particles , and incompressible solvent . this is obviously oversimplified , for we ignore completely any collisions between chains , the deflections caused by weak interactions with the spectator particles , and the fact that there must be some spread in the orientation of the chains themselves @xcite . with these assumptions , in static equilibrium , with no body forces acting , the pressure tensor @xmath1 ( defined as @xmath2 , with @xmath3 the usual stress tensor ) must obey @xmath4 here @xmath5 is an isotropic fluid pressure , and @xmath6 ( @xmath7 ) a compressive stress carried by the force chains . ( [ nn ] ) defines a material that is mechanically very unusual . it permits static equilibrium only so long as the applied compression is along @xmath0 ; while this remains true , incremental loads ( increase or decrease in stresses at fixed major compression of the stress tensor ) can be accommodated reversibly , by what is ( at the particle contact scale ) an elastic mechanism . but the material is certainly not an ordinary elastic body , for if instead one tries to shear the sample in a slightly different direction ( causing a rotation of the principal stress axes ) static equilibrium can not be maintained without changing the director @xmath0 . now , @xmath0 describes the orientation of a set of force chains that pick their ways through a dense sea of spectator particles . accordingly @xmath0 can not simply rotate ; instead , the existing force chains must be abandoned and new ones created with a slightly different orientation . this entails dissipative , plastic , reorganization , as the particles start to move but then re - jam in a configuration that can support the new load . our model jammed colloid is an idealized example of fragile matter " : it can statically support applied shear stresses ( within some range ) , but only by virtue of a self - organized internal structure , whose mechanical properties have evolved directly in response to the load itself . its incremental response can be elastic only to _ compatible _ loads ; _ incompatible _ loads ( in this case , those of a different compression axis ) , even if small , will cause finite , plastic reorganizations . the inability to elastically support _ some _ infinitesimal loads is our proposed definition of the term fragile " ( which , up to now , has not been given a precise technical meaning in this context ) . we argue that jamming may lead _ generically _ to mechanical fragility , at least in systems with overdamped internal dynamics . such a system is likely to arrests as soon as it can support the external load ; since the load is only just supported , one expects the state to be only marginally stable . any incompatible perturbations then force rearrangement ; this will leave the system in a newly jammed but ( by the same argument ) equally fragile state . this scenario is related , but not identical , to several other ideas in the literature @xcite . these include the emergence of rigidity by successive buckling of force chains in glasses and granular matter @xcite ; the concepts of self - organized criticality ( soc ) @xcite ( see also @xcite ) , and those of mechanical percolation which underly recent hypoplastic " models of granular matter @xcite . fragility is also connected with recent ideas concerning isostaticity and marginal mechanics in ( frictionless ) sphere packings @xcite ( see section [ marginal ] ) . our ideas are , in addition , reminiscent of the ( much older ) concept of a self - selecting critical state in theories of soil mechanics @xcite . however the latter is primarily concerned with _ dilatancy _ : the tendency of dense particulate media to expand upon shearing . jamming can be viewed as the constant - volume counterpart of this process : if expansion is prevented , jamming results . consider the idealized jammed colloid of ( fig . [ fig : jammed ] ( a ) ) . so far we allowed for an external stress field ( imposed a the plates ) but no body forces . what body forces can it now support _ without _ plastic rotation of the director ? various models are possible . one is to assume that eq . ( [ nn ] ) continues to apply , with @xmath8 and @xmath9 now varying in space . if @xmath5 is a simple fluid pressure , a localized body force can be supported only if it acts along @xmath0 . thus ( as in a bulk fluid ) no static green function exists for a general body force . ( note that , since eq . ( [ nn ] ) is already written as a continuum equation , such a green function would describes the response to a load that is localized in space but nonetheless acts on many particles in some mesoscopic neighbourhood . ) for example , if the particles in fig . [ fig : jammed ] ( a ) were to become subject to a gravitational force along @xmath10 , then the existing force chains could not sustain this but would reorganize . applying the longitudinal force rule , the new shape is easily found to be a catenary , as realized by hooke @xcite , and emphasized by edwards @xcite . on the other hand , a general body force can be supported , in three dimensions , if there are several different orientations of force chain , possibly forming a network or granular skeleton " @xcite . a minimal model for this is : @xmath11 with @xmath12 directors along three nonparallel populations of force chains ; the @xmath6 s are compressive pressures acting along these . body forces cause @xmath13 to vary in space . we can thus distinguish two levels of fragility , according to whether incompatible loads include localized body forces ( _ bulk _ fragility , _ e.g. _ eq . ( [ nn ] ) ) , or are limited to forces acting at the boundary ( _ boundary _ fragility , _ e.g. _ eq . ( [ osl ] ) ) . in disordered systems one should also distinguish between macro - fragile responses involving changes in the _ mean _ orientation of force chains , and the micro - fragile responses of individual contacts . below we focus on macro - fragility , but in practice the distinction may become blurred . in any case , these various types of fragility should not be associated too strongly with minimal models such as eqs . ( [ nn],[osl ] ) . it is clear that many granular skeletons have a complex network structure where many more than three directions of force chains exist . such a network may nonetheless be fragile ; see section [ marginal ] below . returning to the simple model of eq . ( [ osl ] ) , the chosen values of the three directors ( two in 2-d ) clearly should depend on how the system came to be jammed ( its construction history " ) . if it jammed in response to a constant external stress , switched on suddenly at some earlier time , one can argue that the history is specified purely by the stress tensor itself . in this case , if one director points along the major compression axis then by symmetry any others should lie at rightangles to it ( fig . [ fig : jammed ] ( b ) ) . applying a similar argument to the intermediate axis leads to the ansatz that all three directors lie along principal stress axes ; this is perhaps the simplest model in 3-d . one version of this argument links force chains with the fabric tensor @xcite , which is then taken coaxial with the stress @xcite . with the ansatz of perpendicular directors as just described , eq . ( [ osl ] ) becomes a fixed principle axes " ( fpa ) model @xcite . although grossly oversimplified , this leads to nontrivial predictions for the jammed state in the colloidal problem , such as a constant ratio of the shear and normal stresses when these are varied in the jammed regime . such constancy is indeed reported by laun @xcite in the regime of strong shear thickening " ; see @xcite . we now turn from colloids to granular materials . although the formation of dry granular aggregates under gravity is not normally described in terms of jamming , it is a closely related process . indeed , the filling of silos and the motion of a piston in a cylinder of grains both exhibit jamming and stick - slip phenomena associated with force chains ; see @xcite . and , just as in a jammed colloid , the mechanical integrity of a sandpile disappears as soon as the load ( in this case gravity ) is removed . in the granular context , a model like eq . ( [ osl ] ) is interpreted by saying that a fragile granular skeleton of force chains is laid down at the time when particles are first buried at the free surface ; so long as subsequent loadings are compatible with this structure , the skeleton will remain intact . if in addition the skeleton is rectilinear ( perpendicular directors ) this forces the principal axes to maintain forever the orientation they had close to the free surface ( fpa model ) . however , we do not insist on this last property and other models , based on an oblique family of directors , have also been developed @xcite . ( the construction history of a sandpile allows this since the orientation of the free surface and/or gravity provides a reference direction in addition to that given by the stress tensor at the onset of jamming @xcite . ) note also that , for a conical sandpile we require in addition to eq . ( [ osl ] ) one further relation among stresses , found for example by assuming that @xmath14 everywhere ; see @xcite . it turns out that the fpa models account quite well for the forces measured experimentally beneath conical piles of sand , constructed by pouring cohesionless grains from a point source onto a rough rigid support @xcite . ( note that the two dimensional case a wedge may have special features and the effectiveness of fpa for this is much less clear @xcite . ) as mentioned previously , fragile models such as this show very different mechanics from conventional forms of elasticity or elastoplasticity . for example , in 2-d , when combined with stress continuity ( @xmath15 for sand under gravity ) , eq . ( [ osl ] ) gives differential equations for the stress tensor which are hyperbolic @xcite . with a zero - force boundary condition at the upper surface of a pile , this gives a well - posed problem : the forces acting at the base follow uniquely from the body forces by integration . ( analogous remarks apply to eq . ( [ osl ] ) in 3-d . ) if _ different _ forces are now imposed at the base , rearrangement is inevitable @xcite . ( this is boundary - fragile behaviour . ) the same does not hold @xcite within a traditional elastoplastic modelling approach @xcite whose equations are elliptic in elastic zones . in such models of the sandpile the forces acting at the base can not be found without specifying a displacement field there . to define this displacement , one would normally invoke as _ reference state _ the one in which the load ( gravity ) is removed . for cohesionless poured sand , this state is undefined @xcite , just as it is for a jammed colloid which , in the absence of the applied shear stress , is simply a fluid . one route to an elastic reference state is to consider a hypothetical sandpile where each grain becomes firmly glued " to its neighbours ( or the base ) , upon first coming to rest . the resulting medium is surely elastic , and must therefore be governed by elliptic equations . this does not mean that it is a conventional homogeneous elastic continuum ( for which the states of zero strain and zero stress coincide ) . indeed , a glued pile built under gravity will certainly have nonzero stresses if gravity is later removed @xcite . more importantly , for a typical disordered packing of near - rigid , glued grains , there will arise many _ tensile contacts _ even under a purely compressive external load . thus the problem of glued and unglued piles might , in practice , be extremely dissimilar . the problem of cohesionless granular media is a highly nonlinear one , because ( a ) grains are typically very rigid ( on a scale set by the stresses that arise ) , and ( b ) there can be no tensile forces between any grains in the entire system . the first issue is addressed in @xcite where we show that , for some fragile models , there is in fact a smooth crossover to more conventional elastoplastic physics when grains are deformable , with fragility emerging as the limiting behaviour for rigid particles . in this rigid particle limit , where the longitudinal force rule ( lfr ) becomes valid , fragility will be recovered in a granular skeleton of force chains whose coordination number @xmath16 with @xmath17 dimension of space ( e.g. fig . [ fig : jammed ] ( b ) in two dimensions ) . this is the same rule as applies for packings of _ frictionless _ hard spheres , which also obey the lfr not because of force chains , but because there is no friction . indeed , regular packings of frictionless spheres , which show marginal " or isostatic " mechanics have been studied in detail recently @xcite . note that although our idea of fragility is closely related to the concept of isostaticity ( see e.g. @xcite ) , it is apparently not identical ; in our models of fragile materials the isostatic condition applies only to a loadbearing substructure ( the force skeleton ) and not the whole packing . it would be interesting to test this idea numerically by measuring the mean coordination number among the subset of particles that carry strong forces . interestingly , moukarzel has recently argued @xcite that random packings of frictionless spheres _ generically _ become isostatic ( @xmath16 ) in the rigid particle limit . his arguments appear to depend only on the lfr and the absence of tensile forces , so they should , if correct , equally apply to any granular skeleton that is made of force chains of rigid particles . it remains to be seen whether moukarzel s ( somewhat unintuitive ) arguments can be made rigorous . pending this , we prefer for the moment our own qualitative reasoning , which generically links fragility to jamming , as the motivation for setting @xmath16 in our simplified models ( such as fpa ) of the granular skeleton . such models , in which the skeleton is represented as a rectilinear or oblique array with no disorder , are obviously convenient for calculation . however , disorder will not remove the fragility ( though it will change eqs . ( [ nn],[osl ] ) ) unless it causes the mean coordination number of the skeleton to increase . the jammed state of colloids , if it indeed exists in the laboratory , has not yet been fully elucidated by experiment . it is interesting the even very simple models such as eq . ( [ nn ] ) can lead to nontrivial and testable predictions ( such as the constancy of certain measured stress ratios ) . such models suggest an appealing conceptual link between jamming , force chains , and fragile matter @xcite . however , further experiments are needed to establish the degree to which they are useful in describing real colloids . for granular media , the existence of tenuous force - chain skeletons is clear @xcite ; the question is whether such skeletons are fragile . several theoretical arguments have been given , above and elsewhere , to suggest that this may be the case , at least in the limit of rigid particles . moreover , simulations show strong rearrangement under small changes of compression axis ; the skeleton is indeed self - organized " @xcite . experiments also suggest cascades of rearrangement @xcite in response to small disturbances . these findings are consistent with the fragile picture . the mechanics of fragile models such as eqs . ( [ nn],[osl ] ) differ strongly from those of conventional elasticity . for example , if an infinitesimal point force is exerted on top of a layer of sand , we expect the resulting pressure distribution at the base to form an annulus @xcite ; an elastic model would predict the maximum pressure to be directly beneath the applied force . it could be very fruitful to perform new ( but quite simple ) experiments of this kind . we thank r. ball , e. clement , s. edwards , m. evans , p. evesque , p .- g . de gennes , g. gudehus , j. goddard , j. jenkins , d. levine , j. melrose , s. nagel , j. socolar , c. thornton , l. vanel and t. a. witten for discussions . work funded in part by epsrc ( uk ) gr / k56223 and gr / k76733 . bouchaud , m. e. cates , and p. claudin , j. physique i , * 5 * 639 ( 1995 ) ; j .- bouchaud , p. claudin , m. e. cates , j. p. wittmer , in _ physics of dry granular media _ , h.j . herrmann , j.p . hovi and s. luding , eds . , nato asi 1997 ( cond - mat/9711135 ) . r. s. farr , j. r. melrose , and r. c. ball , phys . e , _ 55 _ , 7203 ( 1997 ) ; r. c. ball , j. r. melrose , adv . colloid interface sci . , _ 59 _ , 19 ( 1995 ) ; j. r. melrose , r. c. ball , europhys . lett . , _ 32 _ , 535 ( 1995 ) . s. alexander , phys . reports * 296 * , 65 ( 1998 ) ; p. s. cundall and o. d. l. strack , in _ mechanics of granular materials : new models and constitutive relations _ , j .. t. jenkins and m. satake , eds , elsevier 1983 . r. hooke , _ a description of helioscopes , and some other instruments _ john martin , london ( 1676 ) . see anagram on title page concerning the true mathematical and mechanical form of all manner of arches . " ; reproduced as fig.2.2 of @xcite . see , e.g. , _ proc . 3rd int . conf . on powders and grains , durham nc 18 - 23 may 1997 _ , r. p. behringer and j. t. jenkins , eds . balkema , rotterdam 1997 , especially f. cantelaube and j. d. goddard , pp . 185 - 194 .	we present simple models of particulate materials whose mechanical integrity arises from a jamming process . we argue that such media are generically fragile " , that is , they are unable to support certain types of incremental loading without plastic rearrangement . in such models , fragility is naturally linked to the marginal stability of force chain networks ( granular skeletons ) within the material . fragile matter exhibits novel mechanical responses that may be relevant to both jammed colloids and cohesionless assemblies of poured , rigid grains . [ multiblock footnote omitted ]
systolic inequalities for surfaces compare length and area , and can therefore be thought of as `` opposite '' isoperimetric inequalities . the study of such inequalities was initiated by c. loewner in 1949 when he proved his torus inequality for @xmath1 ( see pu @xcite and horowitz _ et al . _ the systole , denoted `` sys '' , of a space is the least length of a loop which can not be contracted to a point in the space , and is therefore a natural generalisation of the _ girth _ invariant of graphs . every metric on the @xmath0-dimensional torus @xmath2 satisfies the bound @xmath3 in higher dimensions , m. gromov s deep result @xcite , relying on filling invariants , exhibits a universal upper bound for the systole in terms of the total volume of an essential manifold . l. guth @xcite recently found an alternative proof not relying on filling invariants , and giving a generalisation of gromov s inequality , see also ambrosio and katz @xcite . in dimension @xmath0 , the focus has been , on the one hand , on obtaining near - optimal asymptotic results in terms of the genus @xcite , and on the other , on obtaining realistic estimates in cases of low genus @xcite . one goal has been to determine whether all aspherical surfaces satisfy loewner s bound , a question that is still open in general . we resolved it in the affirmative for genus @xmath0 in @xcite . an optimal inequality of c. bavard @xcite for the klein bottle @xmath4 is stronger than loewner s bound : @xmath5 gromov proved a general estimate for all aspherical surfaces : @xmath6 see ( * ? ? ? * corollary 5.2.b ) . as gromov points out in @xcite , the @xmath7 bound is actually _ optimal _ in the class of finsler metrics . therefore any further improvement is not likely to result from a mere application of the coarea formula . one can legitimately ask whether any improvement is in fact possible , of course in the framework of riemannian metrics . the goal of the present article is to furnish such an improvement in the case of non - orientable surfaces . we will say that such a surface is hyperelliptic if its orientable double cover is . [ theo : main ] let @xmath8 . every riemannian metric from a hyperelliptic conformal type on the surface @xmath9 satisfies the bound @xmath10 note the absence of an ellipsis following `` 333 '' , making our estimate an improvement on gromov s @xmath7 bound . by keeping track of the best constants in our estimates throughout the proof of the theorem , one could obtain a slightly better bound . however , our goal is merely to develop techniques sufficient to improve the @xmath7 bound . since every conformal class on dyck s surface @xmath11 of euler characteristic @xmath12 is hyperelliptic , we have every riemannian metric on dyck s surface @xmath13 satisfies @xmath14 in @xcite , the following result was proved . [ s>3 ] every metric in the conformal class of a hyperelliptic surface @xmath15 of genus @xmath16 satisfies the estimate @xmath17 . given a non - orientable surface with hyperelliptic double cover , we apply proposition [ s>3 ] to conclude that the bound is satisfied whenever @xmath18 . it remains therefore to handle the case @xmath19 . the proof of the main theorem exploits a variety of techniques ranging from hyperellipticity to the coarea formula and cutting and pasting . the current best upper bound for the systole on dyck s surface only differs by about @xmath20 from the best known example given by a suitable extremal hyperbolic surface ( see section [ fourteen ] ) . we will exploit the following characterisation of the systole of a non - orientable surface . given a metric on a klein surface @xmath21 where @xmath22 is fixed point - free , we consider the natural @xmath23-invariant pullback metric on its orientable double cover @xmath24 . the least displacement `` @xmath25 '' of @xmath26 is the number @xmath27 the systole of a klein surface @xmath28 can be expressed as the least of the following two quantities : @xmath29 indeed , lifting a systolic loop of @xmath28 to @xmath24 , we obtain either a loop in the orientable cover @xmath15 , or a path connecting two points which form an orbit of the deck transformation @xmath23 . recall that the following four properties of a closed surface @xmath30 are equivalent : ( 1 ) @xmath30 is aspherical ; ( 2 ) the fundamental group of @xmath30 is infinite ; ( 3 ) the euler characteristic of @xmath30 is non - positive ; ( 4 ) @xmath30 is not homeomorphic to either @xmath31 or @xmath32 . the following conjecture has been discussed in the systolic literature , see @xcite . every aspherical surface satisfies loewner s bound @xmath33 m. gromov @xcite proved an asymptotic estimate which implies that every orientable surface of genus greater than @xmath34 satisfies loewner s bound . this was extended to orientable surfaces of genus at least @xmath35 in @xcite , and for the genus @xmath0 surface in @xcite . recent publications in systolic geometry include ambrosio & katz @xcite , babenko & balacheff @xcite , balacheff _ et al . _ @xcite , belolipetsky @xcite , el mir @xcite , fetaya @xcite , katz _ et al . _ @xcite , makover & mcgowan @xcite , parlier @xcite , ryu @xcite , and sabourau @xcite . in section [ six ] , we will review the relevant conformal information , including hyperellipticity . in section [ two ] , we will present metric information in the context of a configuration of the five surfaces appearing in our main argument . section [ fourteen ] exploits optimal inequalities of blatter and sakai for the mbius band so to prove our first theorem for @xmath19 . we handle the remaining case , namely @xmath36 , in section [ sec : last ] . in this section we review the necessary pre - metric ( i.e. , conformal ) information . the quadratic equation @xmath37 over @xmath38 is well known to possess two distinct solutions for every @xmath39 , and a unique solution for @xmath40 . now consider the locus ( solution set ) of the equation @xmath41 for @xmath42 and generic @xmath43 of even degree @xmath44 . such a locus defines a riemann surface which is a branched two - sheeted cover of @xmath38 . the cover is obtained by projection to the @xmath45-coordinate . the branching locus corresponds to the roots of @xmath43 . there exists a unique smooth closed riemann surface @xmath15 naturally associated with , sometimes called the _ smooth completion _ of the affine surface , together with a holomorphic map @xmath46 extending the projection to the @xmath45-coordinate . by the riemann - hurwitz formula , the genus of the smooth completion is @xmath16 , where @xmath47 . all such surfaces are hyperelliptic by construction . the hyperelliptic involution @xmath48 flips the two sheets of the double cover of @xmath31 and has exactly @xmath44 fixed points , called the weierstrass points of @xmath15 . the hyperelliptic involution is unique . the involution @xmath49 can be identified with the nontrivial element in the center of the ( finite ) automorphism group of @xmath28 ( @xcite ) when it exists , and then such a @xmath49 is unique @xcite . a hyperelliptic closed riemann surface @xmath15 admitting an orientation - reversing ( antiholomorphic ) involution @xmath23 can always be reduced to the form where @xmath43 is a polynomial all of whose coefficients are real , where the involution @xmath50 restricts to complex conjugation on the affine part of the surface in @xmath51 , namely @xmath52 the special case of a fixed point - free involution @xmath23 can be represented as the locus of the equation @xmath53 where @xmath54 for all @xmath55 . here the minus sign on the left hand side ensures the absence of real solutions , and therefore the fixed point - freedom of @xmath23 . the uniqueness of the hyperelliptic involution implies the following . [ comm ] we have the commutation relation @xmath56 . a klein surface @xmath28 is a non - orientable closed surface . such a surface can be thought of as an antipodal quotient @xmath57 of an orientable surface by a fixed point - free , orientation - reversing involution @xmath23 . the pair @xmath58 is known as a real riemann surface . a klein surface is homeomorphic to the connected sum @xmath59 of @xmath60 copies of the real projective plane . the case @xmath61 corresponds to the klein bottle @xmath62 , covered by the torus . in the case @xmath19 , we obtain the surface @xmath63 of euler characteristic @xmath12 , whose orientable double cover is the genus @xmath0 surface @xmath64 . in the sequel , we will focus on dyck s surface @xmath13 since the proof of the main theorem in this case requires special arguments . we will denote by @xmath24 the orientable double cover of a klein surface @xmath28 . if a surface @xmath28 is hyperelliptic , we will denote by @xmath65 its branch locus , i.e. , the set of isolated branch points of the double cover @xmath66 . a klein bottle @xmath4 is called a _ companion _ of a klein surface @xmath28 if we have the inclusion of the branch loci at the level of the orientable double covers : @xmath67 [ com ] each klein surface @xmath4 homeomorphic to @xmath68 admits a triplet of companion klein bottles @xmath69 satisfying the following three conditions : 1 . @xmath70 ; 2 . @xmath71 for @xmath72 ; 3 . @xmath73 . given a real riemann surface @xmath74 , consider the presentation with @xmath43 a real polynomial . we can write the roots of @xmath75 as a collection of conjugate pairs @xmath76 . thus in the genus @xmath0 case , the affine form of the surface is the locus of the equation @xmath77 in @xmath51 . choosing two conjugate pairs , for instance @xmath78 , we can construct a companion surface @xmath79 by the riemann - hurwitz formula , its genus is one , and therefore the ( smooth completion of the ) companion surface is a torus . we will denote it @xmath80 . by construction , its set of zeros is @xmath23-invariant . in other words , the ( affine part in @xmath51 of the ) torus is invariant under the action of complex conjugation . thus , the surface @xmath81 is a klein bottle @xmath4 , namely , a companion klein bottle of the original klein surface @xmath82 stemming from @xmath83 . we thus obtain the three klein bottles @xmath81 , @xmath84 , and @xmath85 , proving the proposition . the maps constructed so far can be represented by the following diagram of homomorphisms ( note that two out of the four arrows point in the leftward direction ) : @xmath86 complex conjugation @xmath23 on @xmath87 fixes a circle called the equatorial circle , denoted @xmath88 . the _ equator _ of @xmath64 is the circle given by the inverse image of the equator @xmath89 under the hyperelliptic projection @xmath90 . the equator of @xmath64 coincides with the fixed point set of the composition @xmath91 . the equator is invariant under the action of @xmath23 . the action of @xmath23 on the equator of @xmath64 is fixed point - free . the lemma is immediate from proposition [ comm ] . similarly , we obtain the following . [ lem : circle ] relative to the double cover @xmath92 , the inverse image of the equatorial circle @xmath93 is a pair of disjoint circles , the involution @xmath23 acts on the torus by switching the two circles , while the involution @xmath91 fixes both circles pointwise . a loop on a non - orientable surface is called @xmath0-sided if its tubular neighborhood is homeomorphic to an annulus , and @xmath94-sided if its tubular neighborhood is homeomorphic to a mbius band . the following lemma is immediate from the homotopy lifting property . [ ed ] let @xmath28 be a klein surface , and @xmath24 its orientable double cover . let @xmath95 be a one - sided loop , and @xmath96 a @xmath0-sided loop in @xmath97 . we have the following four properties : 1 . lifting @xmath98 to @xmath24 yields a path on @xmath24 connecting a pair of points which form an orbit of the involution @xmath23 ; 2 . lifting @xmath99 to @xmath24 yields a closed curve on @xmath28 ; 3 . the inverse image of @xmath98 under the double cover @xmath100 is a circle ( i.e. has a single connected component homeomorphic to a circle ) ; 4 . the inverse image of @xmath99 under the double cover @xmath100 has a pair of connected components ( circles ) . the following blend of topological and hyperelliptic information will be helpful in the sequel . the upperhalf plane in @xmath38 is a fundamental domain for the action of complex conjugation @xmath23 . hence points on the klein surface @xmath101 can be represented by points in the closure of the upperhalf plane . consider the northern hemisphere @xmath102 with the equator included . we will think of the surface @xmath63 as a double cover @xmath103 the double cover is branched along the equator as well as at three additional weierstrass points , corresponding to the points @xmath104 of the usual hyperelliptic cover @xmath105 . we also have an analogue of the hyperelliptic involution , namely the deck transformation @xmath106 fixing the equator and the three weierstrass points . note that the three remaining weierstrass points @xmath107 are mapped to @xmath104 by the involution @xmath23 . a simple loop @xmath108 parallel to the equator decomposes the northern hemisphere into a union of a disk @xmath109 `` north '' of @xmath110 and an annulus @xmath111 `` south '' of @xmath110 : @xmath112 where @xmath113 ( all the branch points of @xmath114 are in the disk @xmath109 ) . the corresponding decomposition of @xmath63 is @xmath115 where @xmath116 is the once - holed torus , and @xmath117 , the mbius band . we will refer to this decomposition as the annulus decomposition , since it is the non - orientable analogue of the decomposition , see figure [ hemisphere ] . in terms of the connected sum notation , the decomposition corresponds to the topological decomposition @xmath118 . ( 0,0)(0,0 ) ( 140,-120)@xmath110 ( 0,-60)@xmath119 ( 8,-55)@xmath120 ( -20,-75)@xmath119 ( -32,-80)@xmath121 ( 30,-70)@xmath119 ( 38,-75)@xmath122 the main idea in the proof of the main theorem is to reduce the general situation to a case of a _ metrically controlled _ annulus decomposition of the klein surface , as explained in proposition [ 141 ] below . a noncontractible loop on @xmath63 satisfying bavard s bound will be called a bavard loop . [ 141 ] the klein surface @xmath123 either contains a bavard loop , or admits a simple loop @xmath124 which separates it as follows : @xmath125 such that moreover 1 . the surface @xmath116 contains the three weierstrass points ; 2 . the mbius band @xmath117 contains the equatorial circle ; 3 . the loop @xmath124 is @xmath49-invariant ; 4 . the loop @xmath99 is of length at most @xmath126 ; 5 . the loop @xmath99 double - covers a loop @xmath127 which lifts to a systolic loop on a companion torus . the proposition will be proved in this section . we start with some preliminary observations . pulling back the metric to a double cover results in a metric of twice the area . we therefore obtain the following lemma . given a @xmath49-invariant metric on the klein surface @xmath63 , we have the following relations among the areas of the surfaces appearing in diagram : @xmath128 as well as the relation @xmath129 . next , we show that averaging the metric improves the systolic ratio . [ 32 ] let @xmath30 be a hyperelliptic orientable double cover of a klein surface @xmath28 . consider a riemannian metric on @xmath30 . then 1 . averaging its metric by the hyperelliptic involution @xmath49 increases the systolic ratio @xmath130 of @xmath30 ; 2 . averaging its metric by the complex conjugation @xmath23 increases the systolic ratio @xmath130 of @xmath30 ; 3 . the ratio @xmath131 defined in increases under averaging . this point was discussed in detail in @xcite . we summarize the argument as follows . express the metric in terms of a constant curvature metric in its conformal class , by means of a conformal factor @xmath132 . thus , in the case of a torus we obtain a metric @xmath133 at a typical point @xmath75 of the torus , where the function @xmath134 is doubly periodic . we average the factor @xmath132 by the hyperelliptic involution @xmath135 , i.e. , we replace @xmath136 by @xmath137 such averaging preserves the total area of the metric . similarly , it preserves the energy of a curve on the surface . choosing a constant speed parametrisation of a systolic loop for the original metric , we see that its energy is preserved under averaging . hence its length is not decreased by averaging . similar remarks apply in the two remaining cases . given a klein surface @xmath123 , we can similarly average the metric by the hyperelliptic involution . hence we may assume without loss of generality that the metrics on both @xmath63 and @xmath64 are @xmath49-invariant . we now consider a companion klein bottle @xmath4 as in proposition [ com ] . we will seek to transplant short loops from @xmath4 to @xmath28 . a systolic loop on @xmath4 is either a 1-sided loop @xmath98 , or a @xmath0-sided loop @xmath99 ( see lemma [ ed ] ) . now consider the real model of @xmath64 , and its three companion tori of type . for each companion torus , we pass to the quotient klein bottle , and find a systolic loop satisfying bavard s bound . we thus obtain three loops @xmath138 , @xmath139 , and @xmath140 . if all three are two - sided , they lift to loops @xmath141 , @xmath142 , and @xmath143 on the tori . let @xmath144 be the projection of the loop @xmath145 to the sphere , and similarly for @xmath146 and @xmath147 . each loop @xmath148 defines a partition of the 6-point set @xmath149 . assuming the loops @xmath138 , etc . , are @xmath0-sided , build the corresponding loops @xmath144 , etc . consider the three partitions of @xmath149 defined by the three loops @xmath150 . if the three partitions are not identical , then there is a bavard loop on @xmath64 . consider a bavard systolic loop @xmath151 of a klein bottle @xmath4 . consider its lift @xmath152 to the torus , and the projection @xmath148 . if two such loops are non - homotopic in @xmath153 we apply the cut and paste technique of @xcite to rearrange segments of the two loops into a pair of loops that lift to closed paths on the genus @xmath0 surface . one of the lifts is necessarily bavard . [ 36 ] if a systolic loop on a companion klein bottle is @xmath94-sided , then we can transplant it to the klein surface @xmath63 , which therefore satisfies bavard s inequality . given a genus @xmath0 surface , consider a companion klein bottle @xmath154 . consider a systolic loop on @xmath155 . if a systolic loop @xmath98 is one - sided , then @xmath98 lifts to a path connecting a pair of points in an orbit of @xmath23 on the torus . the proof is completed by combining lemma [ path ] and lemma [ loop ] below . [ path ] let @xmath156 be a 1-sided loop , and let @xmath157 be the circle which is the connected double cover of @xmath98 . then there is a pair of real points @xmath158 which decompose @xmath159 into a pair of paths : @xmath160 such that each of the paths @xmath161 , @xmath162 projects to a closed curve @xmath163 , @xmath164 on @xmath165 . note that @xmath166 is invariant under the fixed point - free action of @xmath23 on the torus . the loop @xmath167 projects to a loop denoted @xmath168 under the hyperelliptic quotient @xmath169 . let @xmath88 be the fixed point set of @xmath23 acting on @xmath165 . the connected loop @xmath170 is invariant under complex conjugation . therefore it must meet the equator @xmath171 in a point @xmath172 ( see section [ six ] ) . let @xmath173 note that @xmath174 is a self - intersection point of @xmath175 . it may be helpful to think of @xmath175 as a figure - eight loop , with its center - point @xmath174 on the equator @xmath171 . we now view the original loop @xmath156 as a path starting at the image of the point @xmath75 in @xmath4 . we lift the path @xmath98 to a path @xmath176 joining @xmath75 and @xmath177 . the path @xmath161 projects to half the loop @xmath178 , forming one of the hoops of the figure - eight . if we let @xmath179 , we can write @xmath180 . thus the loop @xmath181 is the union of two loops @xmath182 . the following lemma refers to the five surfaces appearing in diagram . [ loop ] let @xmath183 be the path constructed in lemma [ path ] . consider the loop @xmath184 , and lift it to a path @xmath185 . then the image of @xmath186 under the covering projection @xmath187 is a closed curve . to transplant @xmath161 to the klein surface @xmath63 , note that the path @xmath185 may or may not close up , depending on the position of the third pair @xmath188 of branch points of @xmath189 . if @xmath186 is already a loop , then it projects to a bavard loop on the klein surface @xmath190 . in the remaining case , the path @xmath186 connects a pair of opposite points @xmath191 on the surface @xmath64 , where @xmath192 . therefore @xmath186 projects to a bavard loop in this case , as well . by proposition [ 36 ] , it remains to consider the case when each systolic loop @xmath99 of each of the three companion klein bottles @xmath193 is @xmath0-sided . thus each of these bavard loops @xmath194 lifts to a closed curve @xmath195 on the corresponding torus . let @xmath196 be the corresponding loop on the sphere . [ meet ] if @xmath110 meets the equator , then the original klein surface @xmath63 contains a bavard loop . let @xmath197 . let @xmath198 be the points above it in @xmath64 . we lift the path @xmath110 starting at @xmath174 to a path @xmath199 starting at @xmath75 . if @xmath200 closes up , its projection to @xmath63 is the desired bavard loop . otherwise , the path @xmath200 connects @xmath75 to @xmath177 . in this case as well , the path @xmath200 projects to a bavard loop on @xmath82 . it remains to consider the case when @xmath201 . this corresponds to the annulus decomposition case of proposition [ 141 ] , once we show that the loop is simple in the following lemma . let @xmath195 be a systolic loop on the torus , and let @xmath202 be the corresponding loop on the sphere . assume that @xmath201 . then the loop @xmath110 is simple . by hypothesis , the loop @xmath110 lies in a hemisphere , i.e. , one of the connected components of @xmath203 . the typical case of a non - simple loop to keep in mind is a figure - eight curve . denote by @xmath204 the branch points with respect to which @xmath110 has odd winding number . we will think of the curve @xmath110 as defining a connected graph @xmath205 in a plane . the vertices of the graph are the self - intersection points of @xmath110 . each vertex necessarily has valence @xmath206 . by adding the bounded `` faces '' to the graph , we obtain a `` fat '' graph @xmath207 ( the typical example is the interior of the figure - eight ) . more precisely , the complement @xmath208 has a unique _ unbounded _ connected component , denoted @xmath209 . its complement in the plane , denoted @xmath210 contains both the graph @xmath211 and its bounded `` faces '' . since @xmath110 has odd winding number with respect to the branch points @xmath212 of the double cover @xmath213 , they must both lie inside the connected region @xmath207 : @xmath214 the boundary of @xmath207 can be parametrized by a closed curve @xmath215 , thought of the boundary of the outside component @xmath216 so as to define an orientation on @xmath215 ( in the case of the figure - eight loop , this results in reversing the orientation on one of the hoops of the figure - eight ) . note that @xmath217 is a subgraph . since both branch points lie inside , the loop @xmath215 had odd winding number with respect to each of the points @xmath121 and @xmath122 . hence @xmath215 lifts to a noncontractible loop on the torus . if @xmath110 is not simple , then the boundary of the outside region @xmath218 is not smooth , i.e. , the loop @xmath215 must contain `` corners '' and can therefore be shortened , contradicting the hypothesis that @xmath99 is a systolic loop . in this section , we will prove the main theorem for @xmath19 . [ 15b ] let @xmath220 . the bound @xmath221 is satisfied by every metric on dyck s surface @xmath63 . the orientable double cover of the hyperbolic dyck s surface @xmath13 with the maximal systole was described by parlier @xcite . silhol @xcite identified a presentation of its affine form : @xmath222 see also lelivre & silhol @xcite and gendulphe @xcite . the maximal systole on a hyperbolic dyck s surface @xmath13 is equal to @xmath223 resulting in a systolic ratio of @xmath224 . + to prove theorem [ 15b ] , we use the partition @xmath225 , as constructed in proposition [ 141 ] . since we are studying a scale - invariant systolic ratio , we may and will normalize @xmath63 to unit area : @xmath226 by proposition [ 32 ] , we can assume that the metric on @xmath13 is @xmath49-invariant . by proposition [ 141 ] , our desired bound on the systolic ratio of the klein surface @xmath82 reduces to the case when the bavard loops @xmath99 on the companion klein bottles are 2-sided , but their lifts @xmath195 to the triplet of tori @xmath2 project to loops @xmath227 , where each of the three loops @xmath110 defines the same partition of the set @xmath228 . by lemma [ meet ] , we may assume each @xmath110 lies in the open northern hemisphere @xmath114 , and that its lift to @xmath64 produces a non - closed curve . thus , we may assume that each of the simple loops @xmath229 separates the six points @xmath230 into two triplets @xmath231 and @xmath232 . hence its connected double cover in @xmath64 is isotopic to the equatorial circle @xmath233 . the latter is a double cover of the equator @xmath88 ( see section [ six ] ) . + we consider two cases : 1 . @xmath234 ; 2 . @xmath235 . we will find a short loop on @xmath116 in the case ( 1 ) , and on the mbius band in the case ( 2 ) . we will start with a few preliminary topological results . [ lem : dist ] we have @xmath236 , where @xmath237 is the equator of @xmath238 . in particular , if @xmath239 then @xmath240 every path on @xmath165 connecting a branch point to the equator is double covered by a noncontractible loop in @xmath63 . the lemma follows . we will therefore assume throughout that @xmath240 we consider two cases : 1 . @xmath234 ; 2 . @xmath235 . we will find a short loop on @xmath116 in the case ( 1 ) , and on the mbius band in the case ( 2 ) . to prove theorem [ 15b ] , we need to locate an essential loop of length at most @xmath241 , where @xmath242 if there is no such loop , the distance from each of the three branch points to the equator must be at least @xmath243 . note that @xmath244 . we define @xmath245 as follows . let @xmath246 . define @xmath247 as the preimage of @xmath248 under the ramified cover @xmath249 . observe that @xmath250 is a punctured torus and that @xmath251 by identifying the opposite points on @xmath252 by @xmath23 . [ lem : retrac ] if @xmath253 then every arc of @xmath250 with endpoints in @xmath252 of length less than @xmath254 is homotopically trivial in @xmath255 . consider a length - minimizing arc @xmath120 of @xmath250 with endpoints in @xmath252 which is not homotopic to an arc of @xmath252 keeping its endpoints fixed . the arc @xmath120 is a nonselfintersecting geodesic made of two minimizing segments of the same length meeting at a point @xmath45 with @xmath256 . if @xmath120 and @xmath257 agree ( up to orientation ) then the arc @xmath120 passes through a weierstrass point ( which agrees with @xmath45 ) and so @xmath258 from . we will therefore suppose otherwise . by proposition [ 141 ] , the @xmath49-invariant simple loop @xmath259 of length at most @xmath260 lifts to a @xmath49-invariant simple loop in @xmath250 with the same length . this loop will still be denoted by @xmath99 . by construction , the loops @xmath99 and @xmath252 bound a cylinder in @xmath250 . in particular , the arc @xmath120 intersects @xmath99 at exactly two points by minimality of @xmath120 and @xmath99 , @xcite . these two points decompose @xmath120 into three subarcs @xmath261 , @xmath262 and @xmath263 with @xmath261 lying in @xmath116 , figure [ fig : abc ] , that is , @xmath264 . switching @xmath262 and @xmath263 if necessary , we can assume that @xmath262 is no longer than @xmath263 . ( 0,0)(0,0 ) ( -97,-96)@xmath265 ( -25,-96)@xmath261 ( 34,-96)@xmath266 ( -41,-30)@xmath262 ( -41,-158)@xmath263 ( 50,-158)@xmath267 ( 0,-161)@xmath268 ( 99,-120)@xmath99 ( 130,-120)@xmath252 ( -61,-10)@xmath120 ( 51,-10)@xmath257 and @xmath99,height=226 ] the endpoints of @xmath261 and @xmath266 , where @xmath49 is the hyperelliptic involution on @xmath250 , decompose @xmath99 into four arcs @xmath265 , @xmath269 , @xmath268 and @xmath270 , see figure [ fig : abc ] . since @xmath120 and @xmath99 are length - minimizing in their homotopy classes , the loop @xmath271 is noncontractible in @xmath13 and so of length at least @xmath241 . that is , @xmath272 by construction , the arc @xmath273 with symmetric endpoints induces a noncontractible loop in @xmath13 . indeed , its @xmath274-intersection with the projection of @xmath252 in @xmath13 is nontrivial . since the length of this arc is less or equal to the sum of the lengths of @xmath120 and @xmath268 , we obtain @xmath275 on the other hand , we have @xmath276 thus , from , and , we derive @xmath277 we can now introduce the following definition . let @xmath278 be the @xmath279-neighborhood of the equator in @xmath63 with @xmath280 from lemma [ lem : retrac ] ( under the assumption that @xmath281 ) , the boundary components of @xmath278 are formed of one loop freely homotopic to the equator and possibly several other contractible loops . strictly speaking , we should first replace @xmath279 with a nearby regular value of the distance function from the equator . define now @xmath282 as the union of @xmath278 and the disks bounded by its contractible boundary components . note that @xmath282 is a mbius band which does not contain any weierstrass point from lemma [ lem : dist ] [ 54 ] if @xmath239 then @xmath283 . this area lower bound follows from an estimate due to blatter @xcite and used by sakai @xcite . this is a lower bound for the area of a mbius band in terms of its systole and length of path which is non - homotopic to the boundary , and with endpoints at the boundary circle . the lower bound equals half the area of a spherical belt formed by an @xmath279-neighborhood of the equator of a suitable sphere of constant curvature . here the equator of the suitable sphere has length @xmath284 and its radius is @xmath285 . the antipodal quotient of the spherical belt is a mbius band of systole @xmath241 . let @xmath98 be the subtending angle @xmath98 of the northern half of the belt . then @xmath98 satisfies @xmath286 ( here we use the values @xmath287 and @xmath288 ) . the height function is the moment map ( archimedes s theorem ) , and hence the area of the belt is proportional to @xmath289 . the area of the corresponding region on the unit sphere is @xmath290 . hence the area of the spherical belt is @xmath291 , which after quotienting by the antipodal map yields a lower bound @xmath292 proving the proposition . let us show that the same lower bound holds for the area of the mbius band @xmath117 in @xmath13 . for this purpose , we will need the following result . [ 53b ] if @xmath293 then @xmath294 . let @xmath295 be the distance function from the equator of @xmath238 endowed with the metric inherited from @xmath13 . the equator is at distance at least @xmath243 from each of the three isolated branch points by lemma [ lem : dist ] . each noncontractible connected component of the level curves of @xmath296 at distance at most @xmath297 from the equator is longer than @xmath110 . the coarea inequality gives a lower bound for the area of the strip @xmath298 formed by the noncontractible connected components of the level curves corresponding to distances @xmath299 by construction , this strip is disjoint from the projection of @xmath282 on @xmath238 . if @xmath300 , we obtain the following area lower bound for the strip : @xmath301 now , the strip @xmath298 lifts to a region on @xmath116 of double the area , namely area at least @xmath302 , disjoint from @xmath282 . combined with the lower bound of @xmath303 for the area of @xmath282 as in proposition [ 54 ] , this gives a total lower bound @xmath304 for the area of @xmath63 , which contradicts the original normalisation @xmath305 . [ base ] if @xmath293 then the loop @xmath110 is at distance at least @xmath279 from the equator . in particular , @xmath306 . by contraposition , we consider an arc @xmath98 of length less than @xmath279 connecting @xmath110 to the equator . by lemma [ 53b ] , we have @xmath294 . then the path @xmath307 produces a noncontractible loop on @xmath63 of length less than @xmath308 proving the first statement of the proposition . the second statement follows from proposition [ 54 ] since the mbius band @xmath117 in @xmath63 contains @xmath282 . if @xmath309 then @xmath310 . if the loop @xmath110 does not come within distance @xmath279 of the equator , then the mbius band @xmath311 contains the @xmath279-neighborhood of the equator , and therefore satisfies the area lower bound of proposition [ 54 ] , contradicting our hypothesis . hence the loop @xmath110 necessarily comes within distance @xmath279 of the equator , and the proposition results from proposition [ base ] . the remaining case to consider is the case when @xmath312 and therefore @xmath313 . [ 163 ] if @xmath313 , then @xmath314 . we can now proceed to the proof of theorem [ 15b ] . suppose that @xmath239 . the surface @xmath63 is separated into the union @xmath315 where @xmath313 and @xmath316 from proposition [ base ] . here , the separating loop is isometric to a circle twice as long as @xmath110 , and therefore of radius @xmath317 the area of a hemisphere based on such a circle is @xmath318 by lemma [ 53b ] . attaching the hemisphere to the torus with a disk removed produces a torus of total area at most @xmath319 . applying loewner s bound to the resulting torus , we obtain a systolic loop of square - length at most @xmath320 proving the theorem . in this section , we prove theorem [ theo : main ] in the remaining case @xmath36 . recall that a non - orientable surface is called _ hyperelliptic _ if its orientable double cover is . let @xmath321 . every riemannian metric from a hyperelliptic conformal type on the surface @xmath9 satisfies the bound @xmath322 in particular , @xmath323 let @xmath324 . without loss of generality , we can assume that the hyperelliptic invariant metric on @xmath28 has the property that the area of every disk @xmath325 of radius @xmath326 with @xmath327 satisfies @xmath328 see ( * ? ? ? * lemma 3.5 ) . for @xmath60 even with @xmath329 , the orientable double cover of the surface @xmath330 is a hyperelliptic surface @xmath331 of genus @xmath332 . as in lemma [ lem : circle ] , the preimage of the equatorial circle @xmath333 under the double cover @xmath334 is a pair of disjoint circles . this pair of circles bounds the preimage @xmath335 of the northern hemisphere @xmath248 under the previous double cover . the orientation - reversing involution @xmath23 on @xmath331 switches the two boundary components of @xmath250 . we can obtain @xmath28 from @xmath250 by identifying the pairs of points of @xmath252 corresponding to the orbits of the involution @xmath23 . alternatively , we can define @xmath250 by compactifying the open surface @xmath336 , where @xmath337 is the quotient map induced by @xmath23 . let @xmath98 be a length - minimizing arc of @xmath250 joining the two boundary components of @xmath250 . the hyperelliptic involution @xmath49 takes the endpoints of the @xmath94-chain @xmath338 of @xmath250 to their opposite . as these endpoints lie in @xmath252 , and since @xmath49 and @xmath23 agree on @xmath252 , the previous @xmath94-chain of @xmath250 induces a loop @xmath120 in @xmath28 . this loop is noncontractible in @xmath28 . indeed , let @xmath339 be the complex with fundamental group isomorphic to @xmath340 obtained by collapsing the region of @xmath28 outside a sufficiently small tubular neighborhood of the equator to a point . by construction , the loop @xmath120 of @xmath28 projects to a loop of @xmath339 representing twice a generator of @xmath341 . thus , @xmath120 is noncontractible in @xmath28 and so of length at least @xmath342 . we deduce that the distance between the two boundary components of @xmath250 is at least @xmath343 . similarly , the distance between the weierstrass points of @xmath250 and its boundary components is at least @xmath343 , see lemma [ lem : dist ] . likewise , the distance between any pair of weierstrass points is at least @xmath343 . this shows that the open disks @xmath344 of radius @xmath345 centered at the @xmath346 weierstrass points of @xmath250 and the open @xmath345-tubular neighborhoods , @xmath347 and @xmath348 , of the boundary components of @xmath250 are pairwise disjoint . now , as in lemma [ lem : dist ] , the level curves at distance at most @xmath345 from the boundary components of @xmath250 project to curves that separate the isolated branch points from the equator in @xmath248 . thus , each of these level curves has a noncontractible component in @xmath28 and so is of length at least @xmath342 . from the coarea inequality , we deduce that the area of each tubular neighborhood @xmath349 is at least @xmath350 . since the metric satisfies , the area of @xmath344 is at least @xmath351 . adding these lower bounds , we obtain @xmath352 for @xmath60 odd with @xmath353 , the orientable double cover of @xmath354 is a hyperelliptic surface @xmath355 of genus @xmath346 . as previously , the preimage of the equilatorial circle @xmath333 under the double cover @xmath356 is a single circle . this circle bounds the preimage @xmath250 of the northern hemisphere @xmath248 under the previous double cover . the orientation - reversing involution @xmath23 of @xmath355 takes the points of @xmath252 to their opposite points . we can obtain @xmath28 from @xmath250 by identifying the pairs of opposite points of @xmath252 . as previously , the distances between the weierstrass points of @xmath250 and @xmath252 , and between any pair of weierstrass points are at least @xmath343 . arguing as in the previous case , we deduce from the coarea inequality that the area of the @xmath345-tubular neighborhood of @xmath357 is at least @xmath350 . combined with the estimates on the areas of the disjoint disks of radius @xmath345 centered at the weierstrass points of @xmath250 , we obtain @xmath358 proving the proposition . we are grateful to c. croke for helpful discussions . ambrosio , l. ; katz , m. : flat currents modulo @xmath75 in metric spaces and filling radius inequalities , _ comm . * 86 * ( 2011 ) , no . 3 , 557591 . see http://dx.doi.org/10.4171/cmh/234 and arxiv:1004.1374 bangert , v ; croke , c. ; ivanov , s. ; katz , m. : filling area conjecture and ovalless real hyperelliptic surfaces . _ geometric and functional analysis ( gafa ) _ * 15 * ( 2005 ) no . 3 , 577 - 597 . see arxiv : math.dg/0405583 gromov , m. : systoles and intersystolic inequalities . actes de la table ronde de gomtrie diffrentielle ( luminy , 1992 ) , 291362 , _ smin . _ , * 1 * , soc . france , paris , 1996 . gromov , m. : metric structures for riemannian and non - riemannian spaces . based on the 1981 french original . with appendices by m. katz , p. pansu and s. semmes . translated from the french by sean michael bates . reprint of the 2001 english edition . modern birkhuser classics . birkhuser boston , inc . , boston , ma , 2007 . katz , m. ; schaps , m. ; vishne , u. : logarithmic growth of systole of arithmetic riemann surfaces along congruence subgroups . _ j. differential geom . _ * 76 * ( 2007 ) , no . 3 , 399 - 422 . available at arxiv	given a hyperelliptic klein surface , we construct companion klein bottles , extending our technique of companion tori already exploited by the authors in the genus @xmath0 case . bavard s short loops on such companion surfaces are studied in relation to the original surface so to improve a systolic inequality of gromov s . a basic idea is to use length bounds for loops on a companion klein bottle , and then analyze how curves transplant to the original non - orientable surface . we exploit the real structure on the orientable double cover by applying the coarea inequality to the distance function from the real locus . of particular interest is the case of dyck s surface . we also exploit an optimal systolic bound for the mbius band , due to blatter . supported by the israel science foundation grant 1294/06 ] primary 53c23 ; secondary 30f10 , 58j60
the donaldson - thomas ( dt ) invariants enumerate semistable coherent sheaves on calabi - yau 3-folds , which have drawn much attention recently @xcite . we are in interested in the following two problems in dt theory : * find constraints among dt invariants induced by autoequivalences of the derived category of coherent sheaves , e.g. seidel - thomas twists @xcite . * construct dt type invariants counting b - branes on landau - ginzburg ( lg ) models associated to a superpotential . as for the former problem , there are several predictions in string theory on generating series of dt invariants , e.g. @xmath1-duality conjecture , ooguri - strominger - vafa conjecture @xcite , @xcite . there seem to be mysterious constraints among dt invariants behind such predictions , and we hope to reveal their origins via symmetries in the derived category . we believe that a key step toward this problem is to construct a bridgeland stability condition on the derived category @xcite satisfying a certain symmetric property with respect to the given autoequivalence . indeed a construction of a ( weak ) stability condition which is preserved under the derived dual , together with wall - crossing arguments @xcite , @xcite , play crucial roles in the proof of the rationality of the generating series of rank one dt type invariants counting curves @xcite , @xcite , @xcite . as for the latter problem , in order to define the dt type invariants , we need to fix a stability condition for b - branes on lg models . a desired stability condition should be natural in some sense , so that it is an analogue of gieseker stability on coherent sheaves . in a mathematical term , if the superpotential is given by a homogeneous polynomial @xmath2 , the relevant b - brane category is orlov s triangulated category of graded matrix factorizations @xmath3 @xcite . for instance , suppose that @xmath2 is the defining polynomial of a quintic calabi - yau 3-fold @xmath4 . then a desired bridgeland stability condition on @xmath3 may correspond to the gepner point ( cf . figure [ fig : one ] ) of the stringy k@xmath0hler moduli space of @xmath5 , via mirror symmetry and orlov equivalence @xcite @xmath6 the gepner point is an orbifold point in the stringy k@xmath0hler moduli space of @xmath5 , with the stabilizer group @xmath7 . such an orbifold data may be translated into a certain symmetric property of the corresponding bridgeland stability condition , which we focus and pursue in this paper . now we have observed a common keyword regarding the above two problems , that is _ a bridgeland stability condition with a symmetric property_. the motivating problem of this paper , which is rather ambitious and not able to do at this moment , is to find a natural stability condition on @xmath3 , and apply its symmetric property to obtain non - trivial constraints among dt invariants on @xmath5 , via orlov equivalence ( [ intro : orlov ] ) and wall - crossing arguments @xcite , @xcite . subsection [ subsec : future ] ( iii ) . ) hler moduli space of a quintic 3-fold , width=226 ] as we discussed , we are interested in constructing a natural stability condition on @xmath3 . so far there are few works on this problem , except @xcite , @xcite , @xcite , which will be discussed later . a serious issue is that , since the category @xmath3 is _ not _ a priori constructed as a derived category of some abelian category , it is not clear what is the meaning of ` natural ' . our viewpoint is as follows : rather than constructing a stability condition in terms of graded matrix factorizations , we just extract and formulate the symmetric property of a desired stability condition , and try to find a one satisfying such a property . we formulate it as a _ gepner type _ property , which depends on a choice of a pair of an autoequivalence and a complex number , given simply as follows : a stability condition @xmath8 on a triangulated category @xmath9 is called gepner type with respect to @xmath10 if the following condition holds : @xmath11 obviously any stability condition is gepner type with respect to @xmath12 , k)$ ] for @xmath13 , where @xmath14 $ ] is the @xmath15-times composition of the shift functor @xmath16 $ ] . on the other hand , there are several interesting examples in which the relation ( [ intro : eq : gep ] ) holds with respect to non - trivial pairs @xmath17 . as the name indicates , if @xmath4 is a quintic 3-fold , a gepner type stability condition on @xmath18 with respect to the pair @xmath19 conjecturally corresponds to the gepner point , where @xmath20 is the seidel - thomas twist @xcite associated to @xmath21 . the fraction @xmath22 appears since the five times composition of the autoequivalence @xmath23 becomes the twice shift functor . this fact is best understood in terms of @xmath3 via the equivalence ( [ intro : orlov ] ) , as the autoequivalence @xmath23 corresponds to the grade shift functor on the matrix factorization side . at this moment , it seems to be a difficult problem to construct a stability condition on a quintic 3-fold corresponding to the gepner point : we are not even able to construct a bridgeland stability condition on a quintic 3-fold near the large volume limit point . however there is a plenty of examples of weighted homogeneous polynomials , which are more amenable than quintic polynomials , but enough interesting to study . the goal of this paper is to construct gepner type stability conditions on graded matrix factorizations in some of such interesting cases . in the quintic case , an attempt to construct a gepner point leads to a conjectural stronger version of bogomolov - gieseker inequality among chern characters of stable sheaves on quintic 3-folds . the detail in this case will be discussed in a subsequent paper @xcite . let @xmath2 be a homogeneous element with degree @xmath24 in the weighted polynomial ring @xmath25 , \quad \deg x_i = a_i \in \mathbb{z}\end{aligned}\ ] ] such that @xmath26 has only an isolated singularity at the origin . the triangulated category of graded matrix factorizations of @xmath2 , denoted by @xmath27 , is defined to be the homotopy category of the dg category whose objects consist of data @xmath28 where @xmath29 are graded free @xmath30-modules of finite rank , @xmath31 are homomorphisms of graded @xmath30-modules , @xmath32 is the shift of the grading , satisfying the following condition : @xmath33 in @xcite , orlov proved that the triangulated category @xmath27 is related to the derived category of coherent sheaves on the stack @xmath34 depending on the sign of the gorenstein index @xmath35 let @xmath36 be the autoequivalence of @xmath3 induced by the grade shift functor @xmath37 . we propose the following conjecture on the existence of a gepner type stability condition on @xmath27 : [ conj : gep ] there is a gepner type stability condition @xmath38 with respect to @xmath39 , whose central charge @xmath40 is given by @xmath41 here the @xmath42-action on @xmath43 is induced by the @xmath44-grading on each @xmath45 , and @xmath46 is the supertrace which respects the @xmath47-grading on @xmath43 . similarly to the quintic case , the fraction @xmath48 appears since the @xmath24-times composition of @xmath36 coincides with @xmath49 $ ] . we propose that , if there is a desired stability condition @xmath50 in conjecture [ conj : gep ] , then it may be employed as a ` natural ' stability condition for graded matrix factorizations . there are at least three reasons for this . firstly the gepner type property with respect to @xmath39 resembles the following property of the classical gieseker stability on coherent sheaves : for a polarized variety @xmath51 , a coherent sheaf @xmath52 on @xmath5 is @xmath53-gieseker semistable if and only if @xmath54 is @xmath53-gieseker semistable . in this sense , the desired stability condition @xmath50 seems to be a natural analogue of the gieseker stability on @xmath55 for graded matrix factorizations . secondly , the gepner type property with respect to @xmath39 turns out to be a very strong constraint for the stability conditions . indeed , such a property characterizes the central charge @xmath40 uniquely up to a scalar multiplication . subsection [ subsec : unique1 ] . ) also in some cases , we see that @xmath50 is also unique up to shift ( cf . subsection [ subsec : unique2 ] ) and one may expect that this holds in general . thirdly , if there is such @xmath50 , then the @xmath50-semistable objects have a nice compatibility with the serre functor on @xmath3 . as a result , the moduli space of @xmath50-semistable objects should have good local properties , e.g. smoothness , with a perfect obstruction theory , etc , depending on the given data @xmath56 . subsection [ subsec : future ] ( ii ) . ) therefore , whatever the way @xmath50 is constructed , it seems worth studying moduli spaces of @xmath50-semistable objects , and the enumerative invariants defined by them . before stating our result , we mention the previous beautiful works by takahashi @xcite and kajiura - saito - takahashi @xcite . they study the triangulated category @xmath3 in the following cases : @xmath57 @xcite and @xmath58 , @xmath59 @xcite . in these cases , they show that @xmath3 is equivalent to the derived category of representations of a quiver of ade type . as a result , @xmath3 has only a finite number of indecomposable objects up to shift , which are completely classified . using such a classification , they construct a stability condition on @xmath3 by the assignment of a phase for each indecomposable object . their construction satisfies our gepner type property , so conjecture [ conj : gep ] is proved in these cases . subsection [ subsec : rmk : n=3 ] . ) from our motivation , we are rather interested in a case that there is an infinite number of indecomposable objects up to shift , which are hard to classify , and form non - trivial moduli spaces . for instance if @xmath60 , then the stack ( [ w=0 ] ) is either calabi - yau or general type , and it seems hopeless to construct @xmath50 via classification of indecomposable objects . our main result , formulated as follows , contains such cases : ( cf . propositions [ prop:30 ] , [ prop:2 - 1 ] , [ prop:40 ] , [ prop:3 - 1 ] , [ prop:2 - 2 ] . ) [ thm : main ] conjecture [ conj : gep ] is true if @xmath61 and the stack @xmath5 defined by ( [ w=0 ] ) does not contain stacky points . the assumption that @xmath5 does not contain stacky points means that @xmath5 is indeed a smooth projective variety . the inequality @xmath61 implies that @xmath5 is contained in a calabi - yau manifold of dimension less than or equal to two as a codimension @xmath62 complete intersection of hyperplanes . these conditions are restrictive , and the possible types @xmath63 are completely classified . the classification with @xmath64 and @xmath2 of fermat type is given in table 1 below . [ table ] .possible types in theorem [ thm : main ] [ cols=">,>,>,>,>,>",options="header " , ] our strategy proving theorem [ thm : main ] is as follows : by orlov s theorem @xcite , the condition @xmath60 allows us to describe @xmath3 as a semiorthogonal decomposition @xmath65 where @xmath66 is a fully faithful functor from @xmath18 to @xmath3 , and @xmath67 is a certain exceptional object . we show that there is the heart of a bounded t - structure on @xmath3 , given by the extension closure @xmath68 we describe the central charge @xmath40 in terms of the generators of the heart @xmath69 , and take a suitable tilting @xmath70 of @xmath69 using the description of @xmath40 . we then show that the pair @xmath71 determines a stability condition @xmath50 on @xmath3 . it remains to show that @xmath50 has a desired gepner type property , and we reduce it to showing @xmath50-stability of certain objects in @xmath3 . we show that the above construction works in the situation of theorem [ thm : main ] . however unfortunately , several case by case arguments are involved in the proof , and that prevents us to construct @xmath50 beyond the cases in theorem [ thm : main ] . for instance , they contain the proofs of the inequalities of numerical classes of certain objects in @xmath69 , which are required in proving the axiom of bridgeland stability for @xmath71 . they also contain checking the @xmath50-stability of some objects in @xmath3 , which we use to prove the gepner type property of @xmath50 . the above arguments are in particular hard if the image of @xmath40 is not discrete . in the situation of theorem [ thm : main ] , fortunately , the image of @xmath40 is always discrete and that makes our situation technically rather amenable . for instance the image of @xmath40 is not discrete if @xmath72 , and the case of @xmath73 is not included in table 1 . such a case will be discussed in a subsequent paper @xcite , since it involves a subtle argument due to the above non - discrete issue . among the list in table 1 , the case of @xmath74 seems to be the most interesting case , since we observe a new phenomenon relating graded matrix factorizations and coherent systems on the smooth projective curve @xmath5 . recall that a coherent system on @xmath5 consists of data @xmath75 where @xmath76 is a coherent sheaf on @xmath5 and @xmath77 is a finite dimensional @xmath78-vector space . we show that the heart @xmath69 is equivalent to the abelian category of coherent systems on @xmath5 . in the construction of @xmath50 , we see that a clifford type bound on the dimension of @xmath77 is involved . such a clifford type bound for semistable coherent systems is established by lange - newstead @xcite , and we apply their work . if @xmath5 has a higher dimension , the conjectural construction of @xmath50 would predict a higher dimensional analogue of clifford type bound for semistable coherent systems . in the case of a quintic surface , the detail will be discussed in @xcite . we believe that the work of this paper leads to several interesting directions of the future research . we discuss some of them . * ( i ) descriptions of @xmath50-semistable objects in terms of graded matrix factorizations : * as we discussed in the previous subsection , our construction of @xmath50 relies on orlov s theorem , and it is not intrinsic in terms of graded matrix factorizations . it would be an interesting problem to see what kinds of graded matrix factorizations appear as @xmath50-semistable objects , and compare them with the @xmath79-stability discussed in string theory @xcite . it may involve deeper understanding of orlov equivalence , and we are not even able to give a mathematically rigorous candidate of the description of @xmath50 purely in terms of graded matrix factorizations . * ( ii ) constructing moduli spaces of @xmath50-semistable graded matrix factorizations : * it would be an important problem to construct moduli spaces of @xmath50-semistable graded matrix factorizations , and study their properties . we expect that , using the argument of @xcite , there exist artin stacks of finite type @xmath80 which parameterize @xmath50-(semi)stable graded matrix factorizations @xmath43 with @xmath81 . here @xmath82 is the zero - th hochschild homology group of @xmath3 , studied in @xcite , @xcite , @xcite . on the other hand , we have the following vanishing for @xmath83 \in { { \mathcal{m}}}_g^{\rm{ss}}(\gamma)$ ] : @xmath84 the above vanishing , which is proved in lemma [ lem : important ] , is one of the important properties of gepner type stability conditions . since the space @xmath85 is responsible for the local deformation theory of @xmath43 , the moduli space @xmath86 would have good local properties depending on @xmath56 , e.g. it is smooth if @xmath87 , has a perfect obstruction theory if @xmath88 . such properties would be important in constructing counting invariants of graded matrix factorizations , even in a non - cy3 situation . * ( iii ) dt type invariants counting @xmath50-semistable graded matrix factorizations : * suppose that @xmath3 is a cy3 category . if conjecture [ conj : gep ] is true , then ( ii ) would imply the existence of the invariants @xmath89 which count @xmath50-semistable matrix factorizations @xmath43 with @xmath81 . the above invariants may give dt analogue of the fan - jarvis - ruan - witten theory @xcite in gromov - witten theory . also , the gepner type property of @xmath50 should yield an important identity @xmath90 the above identity , combined with orlov s equivalence and wall - crossing argument @xcite , @xcite , may imply non - trivial constraints among the original sheaf counting dt invariants on @xmath5 induced by the autoequivalence @xmath91 . if the above story works , then it realizes an analogue of calabi - yau / landau - ginzburg correspondence in fjrw theory @xcite . in section [ sec : gep ] , we introduce the notion of gepner type stability conditions and propose a conjecture on the existence of gepner type stability conditions on the triangulated category @xmath3 . we also discuss some examples of our conjecture , and their uniqueness . in section [ sec : t - st ] , we construct the heart @xmath69 of a bounded t - structure on @xmath3 , and describe it in terms of quiver representations or coherent systems . in section [ sec : const ] , we explain how to compute the central charge in terms of generators of @xmath69 , and propose a general recipe on a construction of a gepner type stability condition . in section [ sec : proof ] , we prove theorem [ thm : main ] by applying the strategy in section [ sec : const ] . the author would like to thank kentaro hori , kyoji saito and atsushi takahashi for valuable discussions . this work is supported by world premier international research center initiative ( wpi initiative ) , mext , japan . this work is also supported by grant - in aid for scientific research grant ( 22684002 ) from the ministry of education , culture , sports , science and technology , japan . in this section , we recall the definition of bridgeland stability conditions on triangulated categories , group actions , and define the notion of gepner type stability conditions . we then recall orlov s triangulated categories of graded matrix factorizations , and discuss gepner type stability conditions on them . let @xmath9 be a triangulated category and @xmath92 its grothendieck group . we first recall bridgeland s definition of stability conditions on it . [ defi : stab ] _ ( @xcite ) _ a stability condition @xmath8 on @xmath9 consists of a pair @xmath93 @xmath94 where @xmath95 is a group homomorphism ( called central charge ) and @xmath96 is a full subcategory ( called @xmath8-semistable objects with phase @xmath97 ) satisfying the following conditions : * for @xmath98 , we have @xmath99 . * for all @xmath100 , we have @xmath101 $ ] . * for @xmath102 and @xmath103 , we have @xmath104 . * for each @xmath105 , there is a collection of distinguished triangles @xmath106 , \quad e_n = e , \ e_0=0\end{aligned}\ ] ] with @xmath107 and @xmath108 . the full subcategory @xmath109 is shown to be an abelian category , and its simple objects are called @xmath8-stable . in @xcite , bridgeland shows that there is a natural topology on the set of ` good ' stability conditions @xmath110 , and its each connected component has structure of a complex manifold . the above ` good ' conditions are called ` numerical property ' and ` support property ' in literatures . although the above properties are important in considering the space @xmath110 , we omit the detail since we focus on the construction of one specific stability condition . let @xmath111 be the group of autoequivalences on @xmath9 . there is a left @xmath111-action on the set of stability conditions on @xmath9 . for @xmath112 , it acts on the pair ( [ pair2 ] ) as follows : @xmath113 there is also a right @xmath78-action on the set of stability conditions on @xmath9 . for @xmath114 , its acts on the pair ( [ pair2 ] ) as follows : @xmath115 the notion of gepner type stability conditions is defined in terms of the above group actions . a stability condition @xmath8 on a triangulated category @xmath9 is called gepner type with respect to @xmath10 if the following condition holds : @xmath116 here we give some trivial examples : \(i ) for @xmath13 , any stability condition @xmath8 on a triangulated category @xmath9 is gepner type with respect to @xmath12 , k)$ ] . \(ii ) let @xmath117 be an abelian category and @xmath95 a group homomorphism @xmath118 such that @xmath119 is contained in @xmath120 for some @xmath121 . we set @xmath96 for @xmath122 to be @xmath123 other @xmath96 are determined by the rule @xmath101 $ ] . let @xmath23 be an autoequivalence of @xmath124 which preserves @xmath117 . then the pair @xmath125 is a gepner type stability condition on @xmath124 with respect to @xmath126 . a gepner type stability conditions as in example ( ii ) appears at a point in the space of stability conditions on @xmath127 studied by @xcite . such a point corresponds to the orbifold point in the stringy k@xmath0hler moduli space of @xmath5 , and @xmath117 is the abelian category of representations of a mckay quiver . the autoequivalence @xmath23 is given by @xmath128 which induces the automorphism of the mckay quiver . as we will see , there will be more interesting examples of gepner type stability conditions on triangulated categories of graded matrix factorization . here we recall orlov s construction of triangulated categories of graded matrix factorizations @xcite . let @xmath30 be a weighted graded polynomial ring @xmath129 , \quad \deg x_i= a_i\end{aligned}\ ] ] and @xmath130 a homogeneous element of degree @xmath24 . throughout of this paper , we always assume that @xmath131 , and @xmath26 has an isolated singularity at the origin . for a graded @xmath30-module @xmath132 , we denote by @xmath133 its degree @xmath134-part , and @xmath135 the graded @xmath30-module whose grade is shifted by @xmath15 , i.e. @xmath136 . a graded matrix factorization of @xmath2 is data @xmath137 where @xmath29 are graded free @xmath30-modules of finite rank , @xmath31 are homomorphisms of graded @xmath30-modules , satisfying the following conditions : @xmath138 the category @xmath3 is defined to be the homotopy category of graded matrix factorizations of @xmath2 . its objects consist of data ( [ mf ] ) , and the set of morphisms are given by the commutative diagrams of graded @xmath30-modules @xmath139^{p^0}\ar[d]^{f^0 } & p^1 \ar[r]^{p^1}\ar[d]^{f^1 } & p^0(d ) \ar[d]^{f^0(d ) } \\ q^0 \ar[r]^{q^0 } & q^1 \ar[r]^{q^1 } & q^0(d ) . } \end{aligned}\ ] ] modulo null - homotopic morphisms : the above diagram is null - homotopic if there are homomorphisms of graded @xmath30-modules @xmath140 satisfying @xmath141 the category @xmath3 has a structure of a triangulated category . the shift functor @xmath16 $ ] sends data ( [ mf ] ) to @xmath142 and the distinguished triangles are defined via the usual mapping cone constructions . the grade shift functor @xmath37 induces the autoequivalence @xmath36 of @xmath3 , which satisfies the following identity : @xmath143 . \end{aligned}\ ] ] there is also a serre functor @xmath144 on @xmath3 , described by @xmath36 , @xmath145 and the _ gorenstein index _ @xmath146 . the number @xmath146 is defined as @xmath147 the serre functor @xmath144 on @xmath3 is given by ( for instance see ( * ? ? ? * theorem 3.8 ) ) @xmath148 . \end{aligned}\ ] ] the above description of the serre functor will be used later in this paper . the triangulated category @xmath3 is known to be equivalent to the derived category of singularities of the hypersurface singularity @xmath26 . this equivalence is often useful in doing some computations of matrix factorizations . let @xmath79 be the graded ring @xmath149 and @xmath150 the abelian category of finitely generated graded @xmath79-modules . we denote by @xmath151 the subcategory of @xmath152 consisting of perfect complexes of @xmath79-modules . the triangulated category of singularities is defined to be the quotient category @xmath153 the following result is proved in @xcite : _ ( ( * ? ? ? * theorem 3.10 ) ) _ there is an equivalence of triangulated categories @xmath154 sending a matrix factorization ( [ mf ] ) to the cokernel of @xmath155 . the cokernel of @xmath155 is easily checked to be annihilated by @xmath2 , so it is @xmath79-module . obviously the equivalence ( [ cok ] ) commutes with grade shift functors on both sides , so we use the same notation @xmath36 for the grade shift functor on @xmath156 . let @xmath157 be the maximal ideal and set the graded @xmath79-module @xmath158 to be @xmath159 . the graded @xmath79-module @xmath158 determines an object in @xmath156 . the matrix factorization given by @xmath160 plays an important role . by ( * corollary 2.7 ) , it is given by the matrix factorization of the form @xmath161 we omit the descriptions of the above morphisms @xmath155 , @xmath162 , as we will not use them . by an abuse of notation , we abbreviate @xmath163 and denote by @xmath158 the matrix factorization given by ( [ cok ] ) . we are interested in constructing a natural stability condition on @xmath3 . we require that such a stability condition is preserved under the grade shift functor @xmath36 in some sense . this might be an analogues property for the classical @xmath53-gieseker stability on a polarized variety @xmath51 : @xmath53-gieseker semistable sheaves are preserved under @xmath164 . because of the identity ( [ taud ] ) , the expected stability condition is not exactly preserved by @xmath36 but the phases of semistable objects should be shifted by @xmath48 . this is nothing but the gepner type property with respect to @xmath39 . there is a natural construction of a central charge @xmath40 which might give a desired stability condition , and already appeared in some articles @xcite , @xcite , @xcite . for a graded matrix factorization ( [ mf ] ) , its image under @xmath40 is symbolically defined by @xmath165 here @xmath42-action is given by the @xmath166-action on @xmath43 induced by the @xmath44-grading on each @xmath29 , and the @xmath46 means the super trace of @xmath42-action which respects the @xmath47-grading of @xmath167 . more precisely , since @xmath29 are free @xmath30-modules of finite rank , they are written as @xmath168 then ( [ zg ] ) is written as @xmath169 [ exam : c0 ] let @xmath170 be the matrix factorization given by ( [ cok ] ) for @xmath171 . by the definition of @xmath40 , we have @xmath172 it is easy to check that ( [ zg ] ) descends to a group homomorphism @xmath173 indeed , @xmath40 is one of the components of the chern character map of @xmath3 constructed in @xcite . remark [ rmk : chern ] . ) as we stated in the introduction , we propose the following conjecture : [ conj2:gep ] there is a gepner type stability condition @xmath38 with respect to @xmath39 , where @xmath40 is given by ( [ zg ] ) . note that the central charge @xmath40 satisfies the condition ( [ eq : gep ] ) with respect to @xmath39 by the construction . the problem is to construct full subcategories @xmath174 satisfying the desired property . [ discrete ] the image of @xmath40 is contained in @xmath175 $ ] , which may or may not be discrete depending on @xmath24 . for instance , it is discrete if @xmath176 , but not so if @xmath72 . as a toy example of conjecture [ conj2:gep ] , let us consider the case of @xmath177 , i.e. @xmath178 , \quad \deg x = a . \end{aligned}\ ] ] the case of @xmath179 is worked out in @xcite . in this case , the triangulated category @xmath3 is equivalent to the derived category of the path algebra of the dynkin quiver of type @xmath180 . as a result , we have a complete classification of indecomposable objects in @xmath3 , given by @xmath181 for @xmath182 and @xmath183 . note that @xmath184 coincides with @xmath158 given by ( [ cok ] ) . we set @xmath185 to be @xmath186 for @xmath100 , we define @xmath174 to be the subcategory consisting of direct sums of objects @xmath187 with @xmath188 . then the pair @xmath189 is shown to be a desired stability condition in conjecture [ conj2:gep ] by @xcite . the case of @xmath190 follows from the following lemma : [ coprime ] let @xmath30 be the graded ring ( [ def : a ] ) , @xmath191 the greatest common divisor of @xmath192 , and set @xmath193 . let @xmath194 be the graded ring defined by @xmath195 , \quad \deg x_i ' = a_i ' . \end{aligned}\ ] ] for a homogeneous element @xmath130 of degree @xmath24 , we regard it as a homogeneous element @xmath196 of degree @xmath197 by the identification @xmath198 . then if conjecture [ conj2:gep ] holds for @xmath199 , then it also holds for @xmath200 . there is an obvious fully - faithful functor as triangulated categories @xmath201 by multiplying @xmath202 to each grading of @xmath194-modules which appear in the lhs . if @xmath203 is the grade shift on @xmath204 and @xmath205 the central charge ( [ zg ] ) for @xmath204 , we have @xmath206 also note that if there is a non - zero morphism of graded @xmath30-modules @xmath207 , then @xmath208 is divisible by @xmath202 . this implies that @xmath3 has the following orthogonal decomposition : @xmath209 suppose that @xmath210 is a gepner type stability condition on @xmath204 with respect to @xmath211 . we set @xmath212 as follows : @xmath213 by ( [ compare ] ) and ( [ decom ] ) , it is easy to check that @xmath214 is a gepner type stability condition on @xmath3 with respect to @xmath39 . combined with the argument for @xmath57 , we obtain the following corollary : [ cor : n=1 ] conjecture [ conj2:gep ] is true if @xmath177 . we discuss conjecture [ conj2:gep ] in some more cases with small @xmath145 . by corollary [ cor : n=1 ] , the next interesting case may be @xmath215 . in this case , the problem is trivial when @xmath59 . indeed it is easy to check that @xmath216 by using the equivalence ( [ cok ] ) . on the other hand , the triangulated category @xmath3 is non - trivial when @xmath217 . in this case , by applying the coordinate change if necessary , we may assume that @xmath218\end{aligned}\ ] ] with @xmath219 and @xmath220 coprime by lemma [ coprime ] . similarly to the case of @xmath177 , it turns out that there is only a finite number of indecomposable objects up to shift in @xmath3 . they consist of the objects @xmath221 , \quad 0\le j \le d-1 , \quad k\in \mathbb{z}\end{aligned}\ ] ] where @xmath158 is given by ( [ cok ] ) . furthermore each indecomposable objects are mutually orthogonal . the above fact can be easily checked , for instance using orlov s theorem @xcite , as given in example [ exam : n=2 ] below . a desired stability condition @xmath50 in conjecture [ conj2:gep ] is constructed as follows : we first choose @xmath222 so that @xmath223 and set @xmath224)= \phi_0 + k + \frac{2j}{a_1+a_2}. \end{aligned}\ ] ] we define @xmath212 to be the subcategory consisting of direct sums of indecomposable objects with @xmath225 . then by the above argument , @xmath226 gives a desired stability condition . as a summary , we have the following : [ thm : n2e ] conjecture [ conj2:gep ] is true if @xmath215 and @xmath227 . the situation drastically changes when @xmath215 and @xmath228 . for instance , let us consider the case @xmath229 , \quad \deg x_i = 1 , \ \end{aligned}\ ] ] even in such a simple case , conjecture [ conj2:gep ] seems to be not obvious , and it requires a deep understanding of the category @xmath3 . the above case is treated in this paper when @xmath230 . the @xmath72 case will be studied in @xcite . suppose that @xmath58 and @xmath59 , the case studied by kajiura - saito - takahashi @xcite . in this case , @xmath2 is classified into the following ade types @xcite : @xmath231 furthermore , the triangulated category @xmath3 is equivalent to the derived category of quiver representations of a dynkin quiver of the corresponding ade type . as a result , the category @xmath3 is shown to have only a finite number of indecomposable objects up to shift , which are completely classified . similar to the case of @xmath177 ( which is also interpreted as an @xmath232-case in the above @xmath58 , @xmath59 list by kn@xmath233rrer periodicity @xcite ) they assign phases to classified indecomposable objects , and prove the following : _ ( ( * ? ? ? * theorem 4.2))_[thm : kst ] conjecture [ conj2:gep ] is true if @xmath58 and @xmath59 . conjecture [ conj2:gep ] in the case of @xmath58 and @xmath60 is not obvious , and a part of this case is treated later in this paper . it is a natural question whether the gepner type property uniquely characterize @xmath50 or not in some sense . as for the central charge , this is true : @xmath40 is characterized by the gepner type property with respect to @xmath39 up to a scalar multiplication . indeed we are only interested in central charges which factors through the chern character map @xmath234 the rhs is the hochschild homology group of @xmath3 , or more precisely of its dg enhancement . a general theory on hochschild homology groups and chern character maps on @xmath3 is available in @xcite , @xcite , @xcite . because @xmath235 is not finitely generated in general , it would be more natural to define the set of central charges on @xmath3 as the dual space @xmath236 , rather than the original one in definition [ defi : stab ] . on the other hand , the autoequivalence @xmath36 on @xmath3 defines the linear isomorphism @xmath237 the above isomorphism induces the isomorphism @xmath238 on the space of central charges @xmath236 . the gepner type property requires the central charge , regarded as an element in @xmath239 , to be an eigenvector with respect to @xmath238 with eigenvalue @xmath42 . the lemma below shows that such an eigenspace is one dimensional . [ lem : hoch ] the eigenspace of @xmath240-action on @xmath241 with eigenvalue @xmath242 is one dimensional , and contained in @xmath243 . we consider the @xmath244-action on @xmath245 , given by @xmath246 by ( * ? ? ? * theorem 2.6.1 ( i ) ) , @xmath241 is given by @xmath247 here @xmath248 is defined by @xmath249/ ( \partial_{x_1}w , \cdots , \partial_{x_n}w ) \right ) dx_1 \wedge \cdots \wedge dx_n\end{aligned}\ ] ] and the space @xmath250 is given by applying the above construction for @xmath251 since @xmath252 $ ] on @xmath3 , we have @xmath253 on @xmath241 . this implies that @xmath240 generates the @xmath254-action on @xmath241 . by ( * theorem 2.6.1 ( ii ) ) , the decomposition ( [ h : decom ] ) coincides with the character decomposition of @xmath241 with respect to the above @xmath254-action . therefore , noting that @xmath255 , the desired eigenspace is one dimensional by @xmath256 by the grading of @xmath241 given in ( * ? ? ? * theorem 2.6.1 ) , the above eigenspace is contained in @xmath243 . [ rmk : chern ] for @xmath257 , the @xmath258-component of @xmath259 in the decomposition ( [ h : decom ] ) coincides with the central charge @xmath40 defined by ( [ zg ] ) . * theorem 3.3.3 ) . ) in particular @xmath40 is given by an element in @xmath236 , which gives a basis of the eigenspace of @xmath238-action with eigenvalue @xmath42 . [ rmk : action ] obviously the set of gepner type stability conditions with respect to @xmath39 is preserved under the natural right action of @xmath78 on @xmath260 . by bridgeland s deformation result ( * theorem 7.1 ) , lemma [ lem : hoch ] implies that the set of such stability conditions forms a discrete subset in the quotient space @xmath261 . there are some cases in which not only @xmath40 but also @xmath50 is characterized by the gepner type property . at least this is the case when all the indecomposable objects should become semistable . we first note the following lemma , which is an important property of gepner type stability conditions : [ lem : important ] suppose that @xmath130 satisfies conjecture [ conj2:gep ] and @xmath189 is a gepner type stability condition with respect to @xmath39 . for @xmath262 with @xmath263 and @xmath13 , suppose that the following inequality holds : @xmath264 then for any @xmath265 , we have @xmath266 . by the serre functor given by ( [ serre ] ) , we have the isomorphism @xmath267 ) & \cong { \mathop{\rm hom}\nolimits}(f_1 , \tau^{-\varepsilon } ( f_2)[n - k-2])^{\vee}.\end{aligned}\ ] ] by the gepner type property with respect to @xmath39 , we have @xmath268 \in { \mathcal{p}}_g \left ( \phi_2 + n - k-2 -\frac{2\varepsilon}{d } \right ) . \end{aligned}\ ] ] by the inequality ( [ ineq : phases ] ) , the phase of @xmath269 is bigger than that of the above object , hence the rhs of ( [ vanish : serre ] ) vanishes . using the above lemma , we show the following proposition : [ prop : unique ] in the same situation of lemma [ lem : important ] , suppose that the following inequality holds : @xmath270 then all the other stability conditions satisfying the conditions in conjecture [ conj2:gep ] are obtained as @xmath271_{\ast}\sigma_g$ ] for @xmath272 . let us take @xmath265 , @xmath263 with @xmath102 . then lemma [ lem : important ] and the assumption ( [ ineq : ep ] ) show that @xmath273 . this implies that any object in @xmath3 whose harder - narasimhan factors are @xmath269 , @xmath274 decomposes into the direct sum of @xmath269 and @xmath274 . repeating this argument , any object @xmath275 decomposes into the direct sum of @xmath50-semistable objects . in particular any non - zero indecomposable object @xmath257 is @xmath50-semistable , whose phase is denoted by @xmath276 . let us fix a non - zero indecomposable object @xmath277 . by the result of ( * theorem 5.16 ) , the objects @xmath278 generate the triangulated category @xmath3 . therefore for any non - zero indecomposable object @xmath257 , there is @xmath279 and @xmath183 such that @xmath280 ) \neq 0 $ ] . this implies that @xmath281 also by the serre functor ( [ serre ] ) , we have @xmath282 , { \mathcal{s}}_w(e ) ) \neq 0 $ ] , which implies that @xmath283 combined with the assumption ( [ ineq : ep ] ) , we obtain @xmath284 . \end{aligned}\ ] ] now suppose that @xmath285 is another stability condition which satisfies the condition in conjecture [ conj2:gep ] . then there is @xmath272 such that the phase of @xmath286 with respect to @xmath287_{\ast}\sigma_g'$ ] coincides with @xmath288 . if @xmath289 is the phase of the indecomposable object @xmath52 with respect to @xmath287_{\ast}\sigma_g'$ ] , then @xmath289 is also contained in the rhs of ( [ ineq : phase ] ) . since both of the central charges of @xmath50 and @xmath287_{\ast}\sigma_g'$ ] are the same @xmath40 , it follows that @xmath290 . therefore @xmath291_{\ast}\sigma_g$ ] follows . the proof of proposition [ prop : unique ] immediately implies the following : in the same situation of proposition [ prop : unique ] , there is a function @xmath292 from the set of indecomposable objects in @xmath3 to real numbers such that @xmath212 consists of direct sums of indecomposable objects @xmath52 with @xmath293 . the inequality ( [ ineq : ep ] ) is satisfied in the cases of corollary [ cor : n=1 ] , theorem [ thm : n2e ] and theorem [ thm : kst ] . in the list of table 1 , it is satisfied except @xmath294 and @xmath295 . if we believe conjecture [ conj2:gep ] , the proof of proposition [ prop : unique ] predicts that @xmath296 for any non - zero indecomposable object @xmath257 as long as the inequality ( [ ineq : ep ] ) is satisfied . this seems to be not an obvious property of graded matrix factorizations . in this section , we construct and study the hearts of bounded t - structures on @xmath3 , via orlov s theorem relating @xmath3 with the derived category of coherent sheaves on @xmath297 . in what follows , we use the same notation in the previous section . in @xcite , orlov proves his famous theorem relating the triangulated category @xmath3 with the derived category of coherent sheaves on the deligne - mumford stack @xmath298 using the notation in subsection [ subsec : relation ] , orlov s theorem is stated in the following way : _ ( ( * ? ? ? * theorem 2.5))_[thm : orlov ] \(i ) if @xmath59 , there is a fully faithful functor @xmath299 such that we have the semiorthogonal decomposition @xmath300 \(ii ) if @xmath60 , there is a fully faithful functor @xmath301 such that we have the semiorthogonal decomposition @xmath302 in particular @xmath303 is an equivalence if @xmath217 . in this paper we deal with the case of @xmath60 , so we only explain the construction of @xmath303 . it is the composition of the following functors : @xmath304 where @xmath305 is the inverse of ( [ cok ] ) , @xmath306 is the natural projection and @xmath307 is defined by @xmath308 the functor @xmath303 is not compatible with grade shift functors . if @xmath309 , i.e. @xmath5 is a calabi - yau stack , then their difference is described by a seidel - thomas twist functor @xcite on @xmath18 @xmath310 for a spherical object @xmath311 , e.g. a line bundle . the following result is suggested by kontsevich and proved in @xcite : _ ( ( * ? ? ? * proposition 5.8))_[prop : grade ] if @xmath217 , the following diagram commutes : @xmath312^{\psi_i } \ar[d]_{f_i } & { \mathrm{hmf}^{\rm{gr}}}(w ) \ar[d]^{\tau } \\ d^b { \mathop{\rm coh}\nolimits}(x ) \ar[r]^{\psi_i } & { \mathrm{hmf}^{\rm{gr}}}(w ) . } \end{aligned}\ ] ] here @xmath313 . in ( * proposition 5.8 ) , a comparison result similar to proposition [ prop : grade ] is obtained also for @xmath314 . we only mention the case of @xmath309 since we only use the result in this case . [ exam : n=2 ] let us consider the case of @xmath215 and @xmath309 . in the same situation as in subsection [ subsec : rmk2 ] , we have @xmath315 \coprod \left [ \mathrm{pt } / \mathbb{z}_{a_2 } \right ] . \end{aligned}\ ] ] here @xmath316 acts on the smooth one point @xmath317 trivially , and @xmath318 $ ] means the quotient stack . therefore we have the orthogonal decomposition @xmath319 for one dimensional @xmath320-representations @xmath321 with weight @xmath322 . by proposition [ prop : grade ] , the equivalence @xmath323 identifies @xmath36 on @xmath3 with @xmath324 . the equivalence @xmath269 transforms @xmath321 in the following way : @xmath325 \mapsto \cdots \mapsto v_2^{a_2 - 1}[1 ] \mapsto v_1^{0}[2 ] . \end{aligned}\ ] ] in particular @xmath3 has the description stated in subsection [ subsec : rmk2 ] . in this subsection , we construct the hearts of bounded t - structures on @xmath3 when @xmath60 . we introduce the following notation : for a triangulated category @xmath9 and a set of objects @xmath326 , we denote by @xmath327 the extension closure of @xmath328 , i.e. the smallest extension - closed subcategory of @xmath9 which contains objects in @xmath328 . the constructions of our hearts are based on the semiorthogonal decomposition ( [ sod ] ) and the following well - known fact : [ lem : t - st ] let @xmath9 be a triangulated category and @xmath329 a semiorthogonal decomposition . suppose that @xmath330 are hearts of bounded t - structures satisfying @xmath331 for @xmath332 . then there is a bounded t - structure on @xmath9 whose heart @xmath333 is given by @xmath334 . the result is obviously reduced to the case of @xmath335 , which is proved in ( * ? ? ? * lemma 2.1 ) . [ rmk : filt ] in the situation of lemma [ lem : t - st ] , any object @xmath336 admits a filtration @xmath337 such that @xmath338 is an object in @xmath339 . [ rmk : ext ] if the abelian category @xmath339 is generated by an exceptional object @xmath340 , then the heart @xmath333 is the extension closure @xmath341 . in this case , the sequence @xmath342 is called an ext - exceptional collection . we have the following proposition : [ prop : t - st ] suppose that @xmath60 . for each @xmath343 , there is a bounded t - structure on @xmath3 whose heart @xmath344 is given by @xmath345 by following orlov s argument in ( * ? ? ? * theorem 2.5 ) , we see the following : there is an admissible subcategory @xmath346 with a semiorthogonal decomposition @xmath347 and an equivalence @xmath348 which identifies the semiorthogonal decomposition ( [ sod2 ] ) with the rhs of ( [ sod ] ) . here @xmath307 is the functor defined by ( [ def : omega ] ) . therefore by lemma [ lem : t - st ] , it is enough to show that @xmath349 for @xmath350 $ ] with @xmath351 . the assertion ( [ vanish1 ] ) is obvious since @xmath158 is a simple object in the heart @xmath352 . as for the assertion ( [ vanish2 ] ) , since we have @xmath353 for @xmath354 , it follows that @xmath355 here @xmath356 is the zero - th cohomology of @xmath357 . since @xmath356 is concentrated on degree @xmath358 parts , and @xmath158 is on degree @xmath359 parts , the vector space ( [ vanish3 ] ) vanishes . in what follows , we always assume that @xmath60 . we only focus on the case @xmath360 in the above proposition : suppose that @xmath60 . we define @xmath361 and @xmath362 , i.e. @xmath363 this subsection is devoted to investigate some objects in @xmath69 , which will be used later . by definition , we call a closed point @xmath364 _ stacky _ if the stabilizer group at @xmath365 is non - trivial . let us describe @xmath366 for a non - stacky point @xmath367 . note that @xmath5 is a closed substack @xmath368\end{aligned}\ ] ] where @xmath166 acts on @xmath245 via weight @xmath192 . hence @xmath364 is represented by a point @xmath369 . we define the graded @xmath79-module @xmath370 to be @xmath371 where @xmath372 is concentrated on degree @xmath322 , and the action of @xmath373 sends @xmath372 to @xmath374 . obviously if @xmath364 is non - stacky , then @xmath375 is a graded @xmath79-module and isomorphic to @xmath370 . the object @xmath366 is obtained by applying the inverse of ( [ cok ] ) to @xmath370 , after regarding it as an object in @xmath156 . using the above description , we have the following lemma : [ lem : tau : ox ] for any non - stacky point @xmath367 , we have the exact sequence in @xmath69 @xmath376 the result obviously follows from the following exact sequence as graded @xmath79-modules @xmath377 next let us consider the following object @xmath378 the above object is described in terms of the generators of @xmath69 . we have the following lemma : [ lem : filt : a ] we have @xmath379 \in { \mathcal{a}}_w$ ] . furthermore there is a filtration in @xmath69 @xmath380\end{aligned}\ ] ] such that the following holds for @xmath381 @xmath382 let @xmath383 be the maximal ideal ( [ mideal ] ) . we have the exact sequence in @xmath384 @xmath385 which implies @xmath379 \cong m(-\varepsilon)$ ] in @xmath156 . let @xmath386 be the ideal generated by monomials with degree greater than or equal to @xmath134 . we have the following filtration in @xmath384 @xmath387 such that the following holds : @xmath388 therefore it is enough to show that @xmath389 is isomorphic to @xmath390 in @xmath384 . since @xmath391 and @xmath392 for @xmath393 and @xmath394 , we have @xmath395 obviously the rhs is isomorphic to @xmath389 as a graded @xmath79-module . the above lemma can be applied to do some computations on the left adjoint of @xmath66 , denoted by @xmath396 we have the following lemma : [ lem : adj ] the object @xmath397 is isomorphic to @xmath398 $ ] . let @xmath399 is the right adjoint of @xmath66 . note that @xmath400 and @xmath399 are related by @xmath401 where @xmath402 $ ] is the serre functor of @xmath18 . by ( [ serre ] ) , we have @xmath403 \\ & \cong { \mathcal{s}}_x^{-1}(\omega_x[1])[n-2 ] \\ & \cong { \mathcal{o}}_x[1 ] . \end{aligned}\ ] ] here the second isomorphism follows from lemma [ lem : filt : a ] . [ rmk : ext : uni ] by the above lemma , it follows that @xmath404 which is one dimensional for @xmath367 . since @xmath405 is indecomposable , the exact sequence ( [ tau : ox ] ) is a unique non - trivial extension . [ rmk : ext1 ] for @xmath354 , suppose that @xmath406 is a graded @xmath79-module . then lemma [ lem : adj ] and the argument in proposition [ prop : t - st ] imply @xmath407 for @xmath408 , we have the corresponding extension in @xmath384 @xmath409 the graded @xmath79-module @xmath410 is described in the following way : as a graded @xmath78-vector space , it is the direct sum @xmath411 , and the action of @xmath412 sends @xmath413 to @xmath414 . in this subsection , we assume that @xmath215 and describe the heart @xmath69 in terms of certain quiver representations . in this case , @xmath5 is a smooth zero dimensional deligne - mumford stack , and @xmath55 is generated by mutually orthogonal exceptional objects . therefore by ( [ soda ] ) , the abelian category @xmath69 is the extension closure of an ext - exceptional collection . we first compute other hom groups between these exceptional objects , and then describe @xmath69 via the ext - quivers with relations . for some technical reason , we assume that @xmath5 does not contain stacky points , so it consists of finite number of smooth points . then , after applying the coordinate change if necessary , @xmath2 is written as @xmath415 for some homogeneous elements @xmath416 such that @xmath417 ( resp . @xmath418 ) does not divide @xmath419 ( resp . @xmath420 ) . the heart @xmath69 is described in the following way : @xmath421 below we calculate the @xmath422 groups between the above generators . [ lem : compute ] for @xmath423 , we have the following : @xmath424 moreover the natural map @xmath425 sends @xmath426 to @xmath427 under the isomorphism ( [ com : ext ] ) . by the same argument of proposition [ prop : t - st ] , we have @xmath428 then ( [ com : ext ] ) is easily obtained by computing the rhs of ( [ hmf = ext ] ) using the resolution : @xmath429 here @xmath430 and @xmath431 given by matrices @xmath432 next we consider the map ( [ natural ] ) . we write @xmath420 , @xmath419 as @xmath433 for homogeneous elements @xmath434 . let @xmath435 be the dual basis of @xmath436 , and we regard them as elements of the rhs of ( [ hmf = ext ] ) for @xmath437 . then @xmath438 is represented by the morphism of complexes @xmath439^{h(a_1 ) } \ar[d]_{g_1 } & r(0 ) \oplus r(a_1-a_2 ) \ar[r]^{}\ar[d]_{\pi_1 } & r(a_1 ) \ar[d ] \\ r(-a_1 ) \oplus r(-a_2 ) \ar[r]^{(x_1 , x_2 ) } & r(0 ) \ar[r ] & 0 . } \end{aligned}\ ] ] similarly @xmath440 is represented by @xmath441^{h(a_2 ) } \ar[d]_{g_2 } & r(a_2 -a_1 ) \oplus r(0 ) \ar[r]^{}\ar[d]_{\pi_2 } & r(a_2 ) \ar[d ] \\ r(-a_1 ) \oplus r(-a_2 ) \ar[r]^{(x_1 , x_2 ) } & r(0 ) \ar[r ] & 0 . } \end{aligned}\ ] ] here @xmath442 are projections onto the @xmath134-th factor , and @xmath443 are given by matrices @xmath444 the image of @xmath445 by the dual of ( [ natural ] ) is computed by composing the above morphisms of complexes . by restricting the map @xmath446 to the second component of @xmath447 , we see that @xmath448 is mapped to @xmath449 , where @xmath450 is the dual basis of @xmath426 . similarly @xmath451 is mapped to @xmath450 . by dualizing , we obtain the result . [ rmk : ext2 ] by ( [ com : ext ] ) and ( [ hmf = ext ] ) , an element @xmath452 determines the extension in @xmath384 @xmath453 the graded @xmath79-module @xmath410 is isomorphic to @xmath454 as a graded @xmath78-vector space , and the action of @xmath373 is given by sending @xmath413 to @xmath455 . next we compute the hom groups between @xmath158 and @xmath366 for closed points @xmath367 . let @xmath370 be the graded @xmath79-module defined by ( [ def : gm ] ) . we have the following lemma : [ lem : compute2 ] suppose that @xmath456 , and @xmath367 is represented by @xmath457 . then we have @xmath458 here @xmath459 and @xmath460 are regarded as elements @xmath461 where @xmath462 . if @xmath463 , the natural map @xmath464 sends the dual basis @xmath465 to @xmath466 under the isomorphisms ( [ com : ext ] ) , ( [ comp2 ] ) . ( here we set @xmath467 if @xmath468 . ) similarly to the proof of lemma [ lem : compute ] , we have the isomorphism @xmath469 applying @xmath470 to the exact sequence ( [ resol ] ) , the rhs of ( [ isom : x ] ) is computed by the @xmath134-th cohomology group of the following complex @xmath471 here @xmath472 , @xmath473 are the substitution of @xmath474 to the matrices ( [ matrix : h ] ) . then ( [ comp2 ] ) easily follows by noting that every non - zero maps in the complex ( [ com : m ] ) are rank one . the image of @xmath465 by the map ( [ natural2 ] ) is computed similarly to ( [ natural ] ) , so we omit the detail . the above computations enable us to describe @xmath69 in terms of quiver representations with relations . here we only discuss the case of @xmath475 . recall that , given a set of objects @xmath342 , the ext - quiver @xmath476 is defined as follows : the set of vertices is @xmath477 and the number of edges from @xmath322 to @xmath478 is the dimension of @xmath479 , which we identify with a basis of @xmath480 . the following lemma may be well - known , but we include the proof later in subsection [ subsec : lem : equiv ] , because of a lack of a reference . [ lem : equivalence ] let @xmath9 be a triangulated category with finite dimensional hom spaces , generated by an ext - exceptional collection @xmath481 . let @xmath117 be the heart of a bounded t - structure on @xmath9 given by the extension closure of all @xmath482 for @xmath483 . suppose that there is a partition @xmath484 such that , by setting @xmath485 , the following condition holds : @xmath486 then @xmath117 is equivalent to the abelian category of @xmath476-representations with relations generated by the images of the following natural maps for all @xmath487 : @xmath488 the following corollary directly follows from lemma [ lem : compute ] , lemma [ lem : compute2 ] and lemma [ lem : equivalence ] . [ cor : quiver ] suppose that @xmath475 and we write @xmath489 then @xmath69 is equivalent to the category of representations of the quiver of the form @xmath490{\bullet}+!d { } : [ r ] \cdots : [ r ] { \bullet}+!d{v^{(1 ) } } : _ { x_2^{(1)}}^{x_1^{(1)}}[rr]{\bullet}+!d{v^{(0 ) } } ( -^{\pi^{(d)}}[ru]{\bullet}+!d{w^{(d ) } } , -^{\pi^{(j)}}[r]{\vdots } , -_{\pi^{(1)}}[rd]{\bullet}+!d{w^{(1 ) } } ) } \end{aligned}\ ] ] with relations given by @xmath491 for all @xmath492 and @xmath493 . the vertex @xmath494 corresponds to @xmath67 and @xmath495 corresponds to @xmath496 . by investigating the filtration ( [ filt : m ] ) , we are able to describe @xmath379 $ ] in terms of a representation of a quiver ( [ quiver ] ) . the following corollary is a straightforward adaptation of corollary [ cor : quiver ] , remark [ rmk : ext1 ] and remark [ rmk : ext2 ] : [ cor : quiver2 ] in the situation of corollary [ cor : quiver ] , the object @xmath379 $ ] in @xmath69 is the representation of the quiver ( [ quiver ] ) given as follows : @xmath497{\bullet}+!d{r_2 } : [ r ] \cdots : [ r ] { \bullet}+!d{r_{d-3 } } : _ { x_2}^{x_1}[rr]{\bullet}+!d{r_{d-2 } } ( -^{\pi^{(d)}}[ru]{\bullet}+!d{\mathbb{c } } , -[r]{\vdots } , -_{\pi^{(1)}}[rd]{\bullet}+!d{\mathbb{c } } ) } \end{aligned}\ ] ] here @xmath498 is the evaluation at @xmath499 . in this subsection , we assume that @xmath500 and describe the heart @xmath69 in terms of coherent systems on @xmath5 . let us recall the definition of coherent systems . a coherent system on a deligne - mumford stack @xmath5 is data @xmath501 where @xmath77 is a finite dimensional @xmath78-vector space , @xmath354 and @xmath502 is a morphism in @xmath55 . the category of coherent systems on @xmath5 is denoted by @xmath503 . the set of morphisms is given by the commutative diagrams in @xmath55 @xmath504^{s } \ar[d ] & f \ar[d ] \\ v'\otimes { \mathcal{o}}_x \ar[r]^{s ' } & f ' . } \end{aligned}\ ] ] obviously @xmath503 is an abelian category . we have the following proposition : [ prop : syst ] suppose that @xmath500 . then we have an equivalence of abelian categories @xmath505 let us take a coherent system @xmath506 on @xmath5 . by lemma [ lem : adj ] , the morphism @xmath502 is regarded as an element @xmath507 ) . \end{aligned}\ ] ] the cone of @xmath508 determines an object in @xmath69 . the correspondence @xmath509 is a functor from @xmath503 to @xmath69 because , as in the proof of proposition [ prop : t - st ] , we have the vanishing @xmath510 . conversely , let us take an object @xmath511 . there is an exact sequence in @xmath69 @xmath512 for a finite dimensional vector space @xmath77 and @xmath354 . by lemma [ lem : adj ] , the extension class @xmath513 of ( [ ex : sys ] ) is regarded as an element @xmath514 the pair @xmath515 determines an object in @xmath503 . the correspondence @xmath516 is a functor from @xmath69 to @xmath503 since @xmath517 is a semiorthogonal decomposition of @xmath3 . obviously we have @xmath518 hence @xmath519 is an equivalence . combined with lemma [ lem : filt : a ] and remark [ rmk : ext1 ] , we immediately obtain the following corollary : [ cor : canonical ] suppose that @xmath500 . then the object @xmath520 \in { \mathcal{a}}_w$ ] is given by @xmath521 \cong \theta \left ( h^0(x , { \mathcal{o}}_x(1 ) ) \otimes { \mathcal{o}}_x \stackrel{s}{\to } { \mathcal{o}}_x(1 ) \right ) . \end{aligned}\ ] ] here @xmath502 is the canonical evaluation morphism . finally in this section , we give a proof of lemma [ lem : equivalence ] . the proof is straightforward , and probably well - known . we recommend the readers to skip this subsection at the first reading . we denote by @xmath522 the set of relations generated by the images of ( [ relation ] ) . let @xmath523 be the category of @xmath476-representations with relation @xmath522 . we divide the proof into three steps . we construct the functor @xmath524 in the following way : for an object @xmath525 , it admits a filtration @xmath526 such that @xmath527 is written as @xmath528 for finite dimensional vector spaces @xmath529 . by the exact sequence @xmath530 we obtain the linear maps @xmath531 for @xmath532 , @xmath533 , which defines the @xmath476-representation @xmath534 . in order to show that the representation @xmath534 satisfies the relation @xmath522 , consider the composition of extension classes of ( [ eq : e ] ) : @xmath535 \to \bigoplus_{j\in p_k } f_j \otimes v_j[2 ] . \end{aligned}\ ] ] the above composition must vanish since it coincides with the composition @xmath536 \to e_{k+1}/e_{k}[1 ] \to e_{k}/e_{k-1}[2]\end{aligned}\ ] ] where the left morphism is the extension class of @xmath537 by applying @xmath538 for @xmath539 to ( [ compose ] ) , taking the adjunction and @xmath540-component for @xmath541 , we see that the map @xmath542 given by the sum of the composition @xmath543 is zero on @xmath544 , where @xmath545 is the image of ( [ relation ] ) restricted to @xmath546-component . this implies that @xmath534 satisfies the relation @xmath522 , hence it is an object in @xmath523 . the correspondence @xmath547 obviously determines a fully faithful functor from @xmath117 to @xmath523 , since @xmath342 is an ext - exceptional collection . it remains to show that @xmath23 is essentially surjective . let us take an object @xmath548 it consists of finite dimensional vector spaces @xmath529 for @xmath549 and linear maps ( [ linear : vv ] ) whose composition ( [ comp : ii ] ) is zero on @xmath544 for @xmath550 . we need to show the existence of @xmath525 so that @xmath551 . by the induction on @xmath552 , we may assume that the assertion holds for @xmath553 . we set full subcategories @xmath554 as follows : @xmath555 let @xmath556 be the ext - quiver for @xmath557 and define the relation @xmath558 by restricting @xmath522 to @xmath556 . the category @xmath559 is naturally considered as a subcategory of @xmath523 , and there is an exact sequence @xmath560 where @xmath561 and @xmath562 is written as @xmath563 here @xmath372 is the simple object in @xmath523 corresponding to the vertex @xmath322 . by the assumption of the induction , there is an object @xmath564 such that @xmath565 . by the exact sequence ( [ ex : w ] ) , it is enough to show that the map @xmath566 induced by @xmath23 is an isomorphism for all @xmath567 . here we have written @xmath523 just as @xmath568 for simplicity . we show that the morphism ( [ ind : ex ] ) is an isomorphism . let us consider the exact sequence in @xmath117 @xmath569 with @xmath570 and @xmath571 is written as @xmath572 by the condition ( [ lem : cond ] ) , for @xmath567 , we see that @xmath573 and there is a natural isomorphism @xmath574 therefore applying @xmath575 to ( [ ex : fin ] ) , we obtain the exact sequence @xmath576 on the other hand , there is an exact sequence in @xmath523 @xmath577 such that @xmath578 . since @xmath579 unless @xmath580 , @xmath581 with @xmath582 , we have @xmath583 for any @xmath567 . therefore we obtain the commutative diagram @xmath584 & { \mathop{\rm ext}\nolimits}^1_{{\mathcal{a}}}(f_j , e ' ) \ar[r ] \ar[d]^{\alpha } & \qquad \bigoplus_{j ' \in p_{l-1 } } { \mathop{\rm ext}\nolimits}^1_{{\mathcal{a}}}(f_j , f_{j ' } ) \otimes v_{j ' } \ar[d]^{\gamma } \\ 0 \ar[r ] & { \mathop{\rm ext}\nolimits}^1_{\rm{rep}}(e_j , w ' ) \ar[r]^{\delta } & \qquad \bigoplus_{j ' \in p_{l-1 } } { \mathop{\rm ext}\nolimits}^1_{\rm{rep}}(e_j , e_{j ' } ) \otimes v_{j'}. } \end{aligned}\ ] ] here @xmath585 is an isomorphism induced by @xmath23 . by the above diagram , it follows that @xmath586 is injective . on the other hand , the composition @xmath587 vanishes , since any object given by an extension class in the lhs satisfies the relation @xmath522 . therefore @xmath586 is surjective , hence an isomorphism . in this section , we propose a general recipe on a construction of a desired gepner type stability condition . we first compute the central charge @xmath40 in terms of generators of @xmath69 , and try to construct @xmath50 via tilting of @xmath69 . in what follows we assume that the stack @xmath5 in ( [ dmw ] ) is a smooth projective variety , i.e. @xmath5 does not contain stacky points . as in the previous section , we denote by @xmath362 the orlov s fully faithful functor from @xmath18 to @xmath3 , which is an equivalence if @xmath309 . in this subsection , we explain how to compute @xmath40 in the case of @xmath309 . if @xmath309 , @xmath5 is a calabi - yau manifold of dimension @xmath588 , and @xmath69 is equivalent to @xmath55 via @xmath66 . let us consider the group homomorphism given by @xmath589 the above group homomorphism is described in terms of chern characters on @xmath590 . indeed , a fundamental theory on hochschild homology groups implies that @xmath66 induces the isomorphism @xmath591 such that the following diagram commutes : ( cf . * section 1 ) ) @xmath592^{\psi } \ar[d]_{{\mathop{\rm ch}\nolimits } } & { \mathrm{hmf}^{\rm{gr}}}(w ) \ar[d]^{{\mathop{\rm ch}\nolimits } } \\ \mathrm{hh}_0(x ) \ar[r]^{\psi_{\ast } } & \mathrm{hh}_0(w ) . } \end{aligned}\ ] ] we also have the hochschild - kostant - rosenberg isomorphism @xmath593 such that its composition with @xmath594 coincides with the classical chern character map ( * ? ? ? * theorem 4.5 ) . since our central charge @xmath40 factors through the chern character map on @xmath3 ( cf . remark [ rmk : chern ] ) and the poincar pairing on @xmath595 is perfect , the group homomorphism ( [ ghomx ] ) is written as @xmath596 for some @xmath597 . here by an abuse of notation , we also denote by @xmath598 the classical chern character of @xmath599 , and by @xmath600 its @xmath601-component . on the other hand , let us consider the autoequivalence @xmath76 of @xmath18 defined by @xmath602 by proposition [ prop : grade ] , the above autoequivalence corresponds to the grade shift functor @xmath36 on @xmath3 via the equivalence @xmath66 . by the riemann - roch theorem , the autoequivalence @xmath76 acts on @xmath259 for @xmath311 in the following way @xmath603 here @xmath53 is the first chern class of @xmath604 . the above action naturally extends to the linear isomorphism on @xmath595 , given by the composition of the matrices @xmath605 here @xmath606 is the @xmath601-component of @xmath607 , and we regard an element in @xmath595 as a column vector . the gepner type property of the central charge @xmath40 is translated into the following linear equation on @xmath608 : @xmath609 by lemma [ lem : hoch ] , the solution space ( [ linear : eq ] ) must be one dimensional , so it determines the group homomorphism ( [ ghomx ] ) uniquely up to a scalar multiplication . in practice , it is more convenient to work with a smaller subspace in @xmath595 . let @xmath610 be the subspace in @xmath595 defined by @xmath611 we have the following lemma : the solution space of ( [ linear : eq ] ) is contained in @xmath610 . let @xmath612 be the orthogonal complement of @xmath610 in @xmath595 with respect to the poincar paring . we have the direct sum decomposition @xmath613 and @xmath614 preserves the above direct summands . since @xmath615 , @xmath614 acts on @xmath616 via multiplication by @xmath617 , which is unipotent . hence all the eigenvectors of the action @xmath614 on @xmath612 have eigenvalue @xmath618 , which implies that the solution space of ( [ linear : eq ] ) is contained in @xmath610 . by the above lemma , it is enough to solve the equation ( [ linear : eq ] ) for @xmath619 . the ambiguity of the scalar multiplication is fixed by the following lemma : [ lem : fix ] we have the equality @xmath620 by lemma [ lem : adj ] , the object @xmath397 is isomorphic to @xmath398 $ ] . since @xmath309 , the functor @xmath66 is an equivalence , hence @xmath621 . it follows that @xmath622 is isomorphic to @xmath623 $ ] . then the equality ( [ int : alpha ] ) follows by applying the homomorphism ( [ ghomx ] ) to @xmath21 , and using the computation in example [ exam : c0 ] . now the @xmath624 are uniquely determined by the equation ( [ linear : eq ] ) and the normalization ( [ int : alpha ] ) . however for our purpose , it is more convenient to consider a different normalization of @xmath625 . namely we write @xmath625 as a multiple of some non - zero complex number and a central charge on @xmath3 whose image of @xmath366 is @xmath626 . this is possible by the following lemma : [ lem : constant ] for any @xmath367 , we have @xmath627 where @xmath628 is given by @xmath629 which satisfies @xmath630 the equality ( [ cw ] ) follows from example [ exam : c0 ] and lemma [ lem : tau : ox ] . the property ( [ thetaw ] ) follows from @xmath631 we summarize the result in this subsection as follows : suppose that @xmath309 . then for @xmath311 , the central charge @xmath632 is written as @xmath633 where @xmath634 satisfies @xmath635 , and it is the unique solution of the linear equation @xmath636 the purpose of this subsection is to reduce the computation of @xmath40 in the case @xmath228 to that of the case @xmath309 . the strategy is to embed @xmath5 into @xmath638-dimensional calabi - yau manifold @xmath639 and relate @xmath40 with the central charge on @xmath639 . we set @xmath640\end{aligned}\ ] ] and consider the element @xmath641 defined by @xmath642 since we assume that @xmath5 does not contain stacky points , the stack @xmath643 also does not contain stacky points . the variety @xmath639 is a projective calabi - yau manifold with dimension @xmath638 , which contains @xmath5 as a zero locus @xmath644 . let @xmath645 be the graded ring @xmath646 . there is a natural push - forward functor @xmath647 by regarding a graded @xmath79 module as a graded @xmath645-module via the surjection @xmath648 . ( cf . @xcite . ) combined with the equivalence ( [ cok ] ) and the functor @xmath649 in ( [ def : psi ] ) , we obtain the diagram @xmath650^{i_{\ast } } & { \mathrm{hmf}^{\rm{gr}}}(\widehat{w } ) \\ d^b { \mathop{\rm coh}\nolimits}(x ) \ar[r]^{i_{\ast } } \ar[u]^{\psi } & d^b { \mathop{\rm coh}\nolimits}(\widehat{x } ) \ar[u]_{\widehat{\psi}}. } \end{aligned}\ ] ] here @xmath651 is the inclusion and @xmath652 is the equivalence , obtained by applying the same construction of @xmath66 to @xmath653 and @xmath639 . we have the following lemma : the diagram ( [ commute ] ) is commutative . the result follows from the definitions of @xmath66 , @xmath652 and the adjunction for @xmath311 @xmath654 the top arrow @xmath655 of the diagram ( [ commute ] ) obviously commutes with grade shift functors on both sides . also by the functoriality of the hochschild homologies , we have the push - forward functor @xmath656 which preserves the one dimensional eigenspaces in lemma [ lem : hoch ] on both sides . by remark [ rmk : chern ] , the composition @xmath657 differs from @xmath40 by a scalar constant , where @xmath658 is the central charge ( [ zg ] ) on @xmath659 applied for @xmath653 . since @xmath660 , it follows that @xmath661 for any @xmath662 by comparing @xmath663 and @xmath664 given in example [ exam : c0 ] . in particular , using the diagram ( [ commute ] ) , it follows that @xmath665 where @xmath628 coincides with the one defined in ( [ cw ] ) . as a summary , we have the following : suppose that @xmath637 . for @xmath311 , the central charge @xmath632 is written as @xmath666 where @xmath667 satisfies @xmath668 and it is the unique solution of the equation ( [ sol : unique ] ) for @xmath639 . note that @xmath669 is computed by the argument in the previous subsection , since @xmath639 is calabi - yau . later we will use the following data : @xmath670 here @xmath364 and @xmath671 is defined by ( [ thetaw ] ) . the relation ( [ zdag : com ] ) is a consequence of the above arguments and the computation in example [ exam : c0 ] . in this subsection , we explain how desired gepner type stability conditions are constructed . we divide the construction into 3-steps : construction of a slope stability on @xmath69 , construction of @xmath50 via tilting of @xmath69 , and checking the gepner type property of @xmath50 . in the next section , we will apply the recipe here in the case of @xmath61 . our first step is to construct an analogue of a slope stability on @xmath69 . this is a map @xmath672 satisfying the weak seesaw property : for any exact sequence @xmath673 in @xmath69 , we have either @xmath674 the above slope function defines the @xmath675-stability in @xmath69 : an object @xmath511 is @xmath675-(semi)stable if for any exact sequence @xmath673 in @xmath69 , we have the inequality @xmath676 we require that @xmath675-stability satisfies the harder - narasimhan property , i.e. for any @xmath511 , there is a filtration in @xmath69 @xmath337 such that each subquotient @xmath677 is @xmath675-semistable with @xmath678 for all @xmath134 . suppose that there is a slope function @xmath675 as above . we define a pair of full subcategories @xmath679 by @xmath680 the existence of harder - narasimhan filtrations in @xmath675-stability implies that @xmath679 is a torsion pair on @xmath69 . ( cf . @xcite . ) we define @xmath70 to be the associated tilting : @xmath681 \rangle_{\rm{ex } } \subset { \mathrm{hmf}^{\rm{gr}}}(w ) . \end{aligned}\ ] ] the category @xmath70 is the heart of a bounded t - structure on @xmath3 . we try to construct a desired stability condition from the heart @xmath70 , by the following : we say that a triple @xmath682 determines a stability condition if the following condition holds : * for any @xmath683 , we have @xmath684 \}. \end{aligned}\ ] ] * any object in @xmath69 admits a harder - narasimhan filtration with respect to @xmath40-stability . here @xmath40-stability and its harder - narasimhan filtrations are defined by the same way as in the @xmath675-stability , by replacing @xmath675 by @xmath685 $ ] . if the triple ( [ triple ] ) determines a stability condition , it associates a pair @xmath686 in the following way : we define @xmath212 for @xmath687 $ ] to be @xmath688 and other @xmath212 are determined by the rule @xmath689.\end{aligned}\ ] ] if @xmath690 , the above construction is nothing but the one given in ( * ? ? ? * proposition 5.3 ) , and the same argument applies to show that ( [ sigma : dag ] ) is a stability condition . below for an interval @xmath691 , we set @xmath692 note that @xmath693)$ ] coincides with @xmath70 by our construction . we require the local finiteness of our stability condition , i.e. for any @xmath694 , the quasi - abelian category @xmath695 is noetherian and artinian for @xmath696 . * definition 5.7 ) . ) it in particular implies that any object @xmath697 admits a jordan - h@xmath233lder filtration . the local finiteness condition holds if the image of @xmath40 is discrete . by remark [ discrete ] , this is always satisfied in the cases studied in the next section , so we will not take care of the local finiteness . in this step , we assume that the triple ( [ triple ] ) determines a stability condition @xmath50 . we expect that @xmath50 is a gepner type stability condition with respect to @xmath39 . to show this , we consider the following stability condition @xmath698 where @xmath699 is given by @xmath700 it is enough to show that ( [ tau-1 ] ) coincides with @xmath50 . this is equivalent to that @xmath701 ) = { \mathcal{p}}_g((\theta , \theta+1])$ ] , or equivalently @xmath702 \right ) . \end{aligned}\ ] ] we show the equality ( [ a = p ] ) by investigating @xmath50-stability of simple objects in @xmath69 . when @xmath215 we have the following lemma : [ lem : fur ] suppose that @xmath215 and the following inequality holds : @xmath703 if @xmath704 , @xmath705 are @xmath50-semistable for all @xmath367 , then the equality ( [ a = p ] ) holds . let @xmath706 be the phase of @xmath405 for @xmath367 and @xmath707 the phase of @xmath158 for @xmath708 . since @xmath405 and @xmath158 for @xmath709 are objects in @xmath69 , and @xmath70 is obtained as a tilting of @xmath69 , the phases @xmath706 , @xmath707 are contained in @xmath710 $ ] for @xmath709 . on the other hand , the condition ( [ condition : n=2 ] ) implies that @xmath711 by comparing ( [ ineqs ] ) with ( [ zdag : com ] ) , we obtain @xmath712 for @xmath709 . we show that ( [ phases ] ) also holds for @xmath713 . by lemma [ lem : filt : a ] , we have @xmath379 \in { \mathcal{a}}_w$ ] . therefore if @xmath714 is @xmath50-stable , then we have either @xmath715 , \quad \phi_{-\varepsilon } \in ( \theta+1 , \theta+2 ] \quad \mbox { or } \\ \label{latter } & \mathbb{c}(-\varepsilon ) \in { \mathcal{a}}_g[2 ] , \quad \phi_{-\varepsilon } \in ( \theta+2 , \theta+3 ] . \end{aligned}\ ] ] in the case of ( [ former ] ) , the equality ( [ phases ] ) also holds for @xmath713 by the inequalities ( [ ineqs ] ) . in the case of ( [ latter ] ) , we need to exclude the case of @xmath716 if this happens , then @xmath717 , and lemma [ lem : filt : a ] and ( [ ineqs ] ) imply that @xmath718 ) \in \ { \mathbb{r}_{>0 } e^{\sqrt{-1 } \pi \phi } : \phi \in ( \theta , \theta+1 ) \}.\end{aligned}\ ] ] this contradicts to that @xmath719 $ ] , hence ( [ phases ] ) also holds for @xmath713 . note that @xmath720 is generated by @xmath704 for all @xmath367 and @xmath158 for @xmath708 , whose phases are given by ( [ phases ] ) . by the inequality ( [ condition : n=2 ] ) , we have @xmath721 for @xmath722 . noting that there is no non - trivial homomorphism from @xmath96 to @xmath723 if @xmath724 , it follows that @xmath725\right ) , \tau(e ) \right)=0\end{aligned}\ ] ] for any @xmath726 . similarly the inequality ( [ condition : n=2 ] ) implies @xmath727 for @xmath394 . it follows that @xmath728\right ) \right)=0\end{aligned}\ ] ] for any @xmath511 . the above vanishing ( [ vvanish:1 ] ) , ( [ vvanish:2 ] ) imply that the rhs of ( [ a = p ] ) is obtained as a tilting of @xmath720 . hence the result follows from lemma [ lem : tilting ] below : we have used the following lemma : [ lem : tilting ] let @xmath9 be a triangulated category , @xmath729 the heart of a bounded t - structure on @xmath9 , and @xmath730 a group homomorphism . suppose that there are torsion pairs @xmath731 , @xmath732 on @xmath117 such that , for @xmath733 \rangle_{\rm{ex}}$ ] the associated tilting , both of the triples @xmath734 determine stability conditions . then @xmath735 . we may assume that @xmath690 . let us take an object @xmath736 . since @xmath737 is a torsion pair , there is an exact sequence in @xmath117 @xmath738 such that @xmath739 and @xmath740 . also since @xmath741 is a torsion pair , objects in @xmath742 are closed under quotients , hence @xmath743 . suppose that @xmath744 . then since ( [ triples ] ) determine stability conditions for @xmath690 , it follows that @xmath745 . hence the condition @xmath740 implies @xmath746 but the condition @xmath747 implies @xmath748 , which is a contradiction . therefore @xmath749 , and @xmath750 follows . similarly @xmath751 also holds , hence @xmath752 holds . because @xmath753 is an orthogonal complement of @xmath754 in @xmath117 , we have @xmath755 , hence @xmath735 holds . next we discuss the case of @xmath58 . in this case , we need to add an extra check of @xmath50-stability . [ lem : con : n=3 ] suppose that @xmath58 and the following inequality holds : @xmath756 suppose furthermore that @xmath704 is @xmath50-stable , @xmath705 are @xmath50-semistable and the following holds for all @xmath367 : @xmath757 then the equality ( [ a = p ] ) holds . by following the proof of lemma [ lem : fur ] , we have the vanishing ( [ vvanish:1 ] ) , ( [ vvanish:2 ] ) for @xmath758 , @xmath759 with @xmath367 . below we show the above vanishing also holds for @xmath760 for any @xmath354 . then the rhs of ( [ a = p ] ) is shown to be a tilting of @xmath720 , hence lemma [ lem : tilting ] is applied to give the result . let us take an object @xmath761 \right)\end{aligned}\ ] ] and let @xmath762 be the right adjoint functor of @xmath66 . we claim that @xmath763 satisfies @xmath764 the above property will be proved in sublemma [ sublem : vanish ] below , and we continue the proof assuming this fact . let us take the distinguished triangle in @xmath3 @xmath765 where @xmath766 satisfies @xmath767 for a coherent sheaf @xmath76 on @xmath5 , we apply @xmath768 to the distinguished triangle ( [ tr : ab ] ) . since @xmath766 is right orthogonal to @xmath769 , and @xmath770 satisfies the condition ( [ vanish : coho ] ) , we have @xmath771 which proves the vanishing ( [ vvanish:2 ] ) for @xmath760 . similarly applying @xmath772 to the triangle ( [ tr : ab ] ) , and noting that @xmath773 by the property ( [ vanish : coho ] ) , we see that the vanishing ( [ vvanish:1 ] ) for @xmath760 is equivalent to @xmath774 to show ( [ vanish : b ] ) , note that the vanishing ( [ vvanish:1 ] ) for @xmath775 , @xmath776 and the triangle ( [ tr : ab ] ) imply @xmath777 therefore , if we denote by @xmath778 the @xmath134-th cohomology with respect to the t - structure on @xmath3 with heart @xmath69 , then we have @xmath779 for @xmath780 . since @xmath781 , this implies that ( [ vanish : b ] ) holds . therefore the vanishing ( [ vvanish:1 ] ) for @xmath760 holds . we have used the following sublemma : [ sublem : vanish ] the condition ( [ vanish : coho ] ) holds . we investigate the vanishing @xmath782 for @xmath367 and @xmath183 . by the serre duality on @xmath5 , adjunction and applying @xmath36 , the vanishing ( [ vanish : inv ] ) is equivalent to @xmath783 , \tau(a))=0 . \end{aligned}\ ] ] we first show that ( [ vanish : inv2 ] ) holds for any @xmath784 and @xmath367 . applying the serre functor @xmath785 $ ] on @xmath3 , the vanishing ( [ vanish : inv2 ] ) is equivalent to @xmath786 , \tau^{1-\varepsilon}\psi({\mathcal{o}}_x))=0 . \end{aligned}\ ] ] on the other hand , by the assumption ( [ condition : n=3 ] ) , we have the inequality @xmath787 for @xmath784 . therefore by our assumptions ( [ check : add ] ) and ( [ cond : a ] ) , the vanishing ( [ vanish : dual ] ) holds for @xmath784 . next , we have the inequality @xmath788 for @xmath722 by ( [ condition : n=3 ] ) , hence the vanishing ( [ vanish : inv2 ] ) holds for @xmath722 . moreover the above inequality , hence the vanishing ( [ vanish : inv2 ] ) , also holds for @xmath789 unless @xmath790 holds . suppose that @xmath790 holds , and let @xmath132 be the @xmath50-harder - narasimhan factor of @xmath791 with the maximum phase . note that @xmath704 is @xmath50-stable with phase @xmath792 , which is bigger than or equal to the phase of @xmath132 . therefore the vanishing of ( [ vanish : inv2 ] ) for @xmath789 does _ not _ hold only if we have @xmath793 and @xmath405 is one of the jordan - h@xmath233lder factors of @xmath132 . it follows that , by taking the jordan - h@xmath233lder filtration of @xmath132 , there is a distinguished triangle @xmath794 where @xmath795 is a zero dimensional coherent sheaf on @xmath5 , @xmath194 is an object in the rhs of ( [ cond : a ] ) , such that the vanishing ( [ vanish : inv2 ] ) holds for @xmath394 after replacing @xmath30 by @xmath194 . applying @xmath796 to ( [ dist : q ] ) , we have the distinguished triangle @xmath797 the above argument shows that , after replacing @xmath30 by @xmath194 , the vanishing ( [ vanish : inv ] ) holds unless @xmath798 . it follows that @xmath799 is a two term complex of vector bundles on @xmath5 , whose cohomologies are concentrated on degrees @xmath800 and @xmath618 . since @xmath801 , we conclude that @xmath802 for @xmath803 . by the above results , the problem is reduced to showing @xmath50-stability of some objects in @xmath3 . the following lemma is useful in checking the @xmath50-stability of these objects . the proof is obvious , and we omit it . [ lem : cmin ] let @xmath9 be a triangulated category , @xmath730 a group homomorphism and @xmath729 the heart of a bounded t - structure on @xmath9 . suppose that the triple @xmath804 determines a stability condition on @xmath9 , and the following condition holds : @xmath805 then an object @xmath525 with @xmath806 is @xmath50-stable if and only if @xmath807 for any @xmath808 with @xmath809 . in this section , we apply the strategy in the previous section and prove theorem [ thm : main ] . below we use the same notation in the previous section . in particular the constants @xmath810 and @xmath671 are defined as in lemma [ lem : constant ] , both in @xmath309 and @xmath637 case . the goal is to prove conjecture [ conj2:gep ] when @xmath811 and @xmath5 does not contain stacky points . since we already discussed the case with @xmath177 and @xmath215 , @xmath309 , the following five possibilities are left : @xmath812 we divide the proof into the five subsections , so that each subsection corresponds to one of the above types . we repeat similar arguments in these subsections , so we recommend the readers to follow only one or two cases , e.g. @xmath294 or @xmath295 , at the first reading of this paper . in this subsection , we study the case of @xmath58 and @xmath309 . in this case , @xmath5 is a smooth elliptic curve , and the heart @xmath69 is given by @xmath813 furthermore possible data @xmath814 are classified into the three types ( * ? ? ? * table 2 ) @xmath815 the central charge @xmath40 is described as follows : note that , since @xmath5 is an elliptic curve , the space of stability conditions on @xmath18 is completely described in ( * ? ? ? * section 9 ) . since @xmath822 , and the central charge @xmath40 is given by ( [ cent : n=3 ] ) , it follows that the triple ( [ triple:30 ] ) satisfies the condition ( [ htheta ] ) . then the same argument of ( * ? ? ? * example 5.4 ) shows that the triple ( [ triple:30 ] ) determines a stability condition @xmath50 on @xmath3 . now we are going to apply lemma [ lem : con : n=3 ] . note that the inequality ( [ condition : n=3 ] ) is satisfied in this case . for a point @xmath367 , we have @xmath823)\end{aligned}\ ] ] by proposition [ prop : grade ] . since the above object is @xmath50-stable with phase @xmath824 , lemma [ lem : con : n=3 ] implies that @xmath50 is a gepner type stability condition with respect to @xmath39 . in this subsection , we study the case of @xmath215 , @xmath500 . in this case , @xmath5 is a finite number of smooth points , and the heart @xmath69 is given by @xmath825 the following lemma immediately follows from ( [ zdag : com ] ) : [ zg:2 - 1 ] if we write the k - theory class of an object @xmath511 as @xmath826 = v_0 [ \mathbb{c}(0 ) ] + \sum_{x\in x } w_x [ \psi({\mathcal{o}}_x)]\end{aligned}\ ] ] and set @xmath827 , then @xmath828 is given by @xmath829 in the notation used in subsection [ subsec : central2 ] , the polynomial @xmath653 is of type @xmath830 , which must belong to one of the classifications in ( [ cla : n=3 ] ) . furthermore , applying lemma [ lem : compute2 ] and lemma [ lem : equivalence ] , we see that @xmath69 is equivalent to the abelian category of representations of a certain quiver @xmath831 . by the above arguments , the possible types of @xmath832 are classified as follows : @xmath833{\bullet}+!d { } , : ^{\pi^{(2)}}[r]{\bullet}+!d { } , : _ { \pi^{(1)}}[rd]{\bullet}+!d { } ) } \\ & ( a_1 , a_2 , d , \sharp x)=(2 , 1 , 4 , 2 ) , \quad { \mathcal{q}}= \xygraph { { \bullet}+!d { } ( : ^{\pi^{(2)}}[ru]{\bullet}+!d { } , : _ { \pi^{(1)}}[rd]{\bullet}+!d { } ) } \\ & ( a_1 , a_2 , d , \sharp x)=(3 , 2 , 6 , 1 ) , \quad { \mathcal{q}}= \xygraph { { \bullet}+!d { } : _ { \pi^{(1)}}[r]{\bullet}+!d { } } \end{aligned}\ ] ] here the left vertex of the quiver @xmath831 corresponds to @xmath170 , and the right vertices correspond to @xmath366 for @xmath367 . in all the above cases , we set @xmath820 so that the heart ( [ ag : tilting ] ) coincides with @xmath69 . we have the following result : by lemma [ zg:2 - 1 ] , it follows that the triple ( [ triple:2 - 1 ] ) satisfies the condition ( [ htheta ] ) . since @xmath69 is noetherian and artinian , the triple ( [ triple:2 - 1 ] ) satisfies the harder - narasimhan property ( cf . * proposition 2.4 ) ) hence it determines a stability condition @xmath50 on @xmath3 . in order to show the gepner type property of @xmath50 , it is enough to check the assumption of lemma [ lem : fur ] . since the inequality ( [ condition : n=2 ] ) is satisfied , it remains to show the @xmath50-stability of @xmath405 and @xmath835 for all @xmath367 , which we prove in the next lemma . as for @xmath704 , it fits into a unique non - trivial extension in @xmath69 by lemma [ lem : tau : ox ] and remark [ rmk : ext : uni ] @xmath836 therefore @xmath366 is the only non - trivial subobject of @xmath704 in @xmath69 . comparing the argument of @xmath837 , we see that @xmath704 is @xmath40-stable . as for @xmath835 , note that the object @xmath520 $ ] is contained in @xmath69 by lemma [ lem : filt : a ] . let @xmath838 be a proper subobject of @xmath520 $ ] in @xmath69 . as in corollary [ cor : quiver2 ] , the inclusion @xmath839 $ ] is represented by the following inclusion of quiver representations : @xmath840{\bullet}+!d{w^{(\sharp x ) } } , -[r]{\vdots}+!d { } , : _ { } [ rd]{\bullet}+!d{w^{(1 ) } } ) } \quad \subset \quad \mathbb{c}(1)[-1]= \xygraph{!~:{@ { > } } ! ~-{@ { } } { \bullet}+!d{r_1 } ( : ^{\pi^{(\sharp x)}}[ru]{\bullet}+!d{\mathbb{c } } , -[r]{\vdots}+!d { } , : _ { \pi^{(1)}}[rd]{\bullet}+!d{\mathbb{c } } ) } \end{aligned}\ ] ] here @xmath841 is the space of degree one elements in @xmath842 $ ] , whose dimension is @xmath843 . the morphisms @xmath844 are evaluations at the points in @xmath5 , and @xmath845 , @xmath846 are finite dimensional vector spaces with @xmath847 , @xmath848 . let @xmath522 be the subset of @xmath849 satisfying @xmath850 . then we have @xmath851 hence @xmath852 . since we have @xmath853 ) & = c_w \left\ { -\sharp x + ( \sharp x -1 ) \cdot ( 1-e^{2\pi \sqrt{-1}/d } ) \right\ } \\ z_g(f ) & = c_w \left\ { \lvert i \rvert - \sharp x + \dim v^{(0 ) } \cdot ( 1-e^{2\pi \sqrt{-1}/d } ) \right\ } \end{aligned}\ ] ] it follows that the argument of @xmath854 is less than that of @xmath855)$ ] in @xmath856 $ ] . therefore @xmath520 $ ] is @xmath50-stable . in this subsection , we study the case of @xmath294 . in this case , @xmath5 is a smooth projective k3 surface , and the heart @xmath69 is given by @xmath858 by theorem [ thm : orlov ] , the triangulated category @xmath3 is equivalent to @xmath18 via the equivalence @xmath66 . on the other hand , the spaces of stability conditions on k3 surfaces are studied by bridgeland @xcite . combining the techniques in @xcite with the arguments in subsection [ subsec : strategy ] , we construct a desired gepner type stability condition on @xmath3 . let us first describe the matrix @xmath286 which appears in the equation ( [ linear : eq ] ) . by writing @xmath859 for @xmath860 , the matrix @xmath286 is given by @xmath861 by using the above description , we give a classification of possible types . by ( [ matrix : m ] ) , we have @xmath864 since @xmath865 for @xmath866 , we have @xmath867 this is only possible for @xmath868 . in the latter case , the possibility @xmath869 is excluded since the stack @xmath5 always contains a stacky point . therefore we obtain the classification ( [ types : n=4 ] ) . in order to describe the central charge @xmath40 , it is convenient to use the twisted mukai vector . for @xmath873 , we set @xmath874 we denote by @xmath875 the @xmath876-component of @xmath877 . the central charge @xmath40 is computed in the following way : by setting @xmath878 , we consider the slope function @xmath675 on @xmath885 , defined by @xmath886 here @xmath887 is defined to be @xmath888 if @xmath52 is a torsion sheaf . the above @xmath675-stability coincides with the classical slope stability condition on @xmath55 via @xmath66 . as we discussed in subsection [ subsec : strategy ] , the slope function @xmath675 defines a torsion pair ( [ mu : tilting ] ) on @xmath69 , and the associated tilting @xmath70 is given by ( [ ag : tilting ] ) . we have the following result : instead of the triple ( [ triple:40 ] ) , we consider the triple @xmath890 , \theta=0).\end{aligned}\ ] ] obviously , the triple ( [ triple:40 ] ) determines a stability condition if and only if the triple ( [ triple:400 ] ) determines a stability condition . if the latter holds , then by remark [ rmk : int:40 ] , the resulting stability condition is @xmath891 of one of stability conditions on @xmath18 constructed in ( ? ? ? * section 6 ) . therefore , applying ( * ? ? ? * lemma 6.2 ) , it is enough to show the following : for any spherical torsion free sheaf @xmath52 on @xmath5 , one has @xmath892 . the proof of this fact requires some more arguments , so we leave the proof to lemma [ lem : spherical ] below . we continue the proof assuming this fact . by the above argument and lemma [ lem : spherical ] , the triple ( [ triple:40 ] ) determines a stability condition @xmath50 on @xmath3 . it remains to show that @xmath50 is gepner type with respect to @xmath39 . to show this , let us consider the objects @xmath893)$ ] for @xmath367 . by proposition [ prop : grade ] , we have @xmath894 ) \cong \psi(i_x ) \in { \mathcal{a}}_g[1 ] \end{aligned}\ ] ] where @xmath895 is the ideal sheaf which defines @xmath365 . in lemma [ lem : ix : stable ] below , we see that @xmath896 is @xmath50-stable , hence it lies in @xmath897 . if we set @xmath285 to be the stability condition defined as in ( [ tau-1 ] ) , this implies that @xmath898 ) \in \tau^{-1 } { \mathcal{p}}_g \left ( \theta_w + \frac{2}{d } \right ) = { \mathcal{p}}_g'(\theta_w)\end{aligned}\ ] ] for any @xmath367 , and it is @xmath285-stable . since @xmath5 is a k3 surface , this implies that @xmath899)$ ] is obtained as a tilting of @xmath900 , by the argument of ( * ? ? ? * lemma 10.1 ) . therefore @xmath901)$ ] coincides with @xmath70 by lemma [ lem : tilting ] , which shows that @xmath50 is gepner type with respect to @xmath39 . this is equivalent to saying that , for any spherical torsion free sheaf @xmath52 on @xmath5 with @xmath902 , one has @xmath903 since @xmath52 is a spherical sheaf , we have @xmath904 , where the square is defined in the mukai lattice of @xmath5 . * lemma 5.1 ) . ) combined with @xmath902 and the hodge index theorem , we have @xmath905 noting that @xmath906 , we obtain @xmath907 , hence @xmath908 the rhs of ( [ ineq : vector ] ) is positive if @xmath909 . therefore we may assume that @xmath910 , i.e. @xmath52 is rank one . since @xmath52 is a spherical sheaf , this implies that @xmath52 is a line bundle . let @xmath552 be the first chern class of @xmath52 . since @xmath911 , we have @xmath912 , and @xmath913 is written as @xmath914 suppose by a contradiction that ( [ spherical ] ) does not hold for a line bundle @xmath52 . then , combining with the hodge index theorem , we obtain @xmath915 since we have only two cases , @xmath916 , and @xmath917 is an even integer , it follows that @xmath918 . also since @xmath919 is either @xmath920 ( when @xmath5 is a quartic k3 surface ) or @xmath618 ( when @xmath5 is a double cover of @xmath872 ) it follows that the linear system @xmath921 defines an elliptic fibration @xmath922 . in particular , its general fiber is a smooth elliptic curve @xmath923 in @xmath5 . however this is a contradiction : when @xmath5 is a quartic k3 surface , we have the castelnuovo inequality @xmath924 which contradicts to @xmath925 . when @xmath5 is a double cover of @xmath872 , if @xmath926 is the double cover , then @xmath927 implies that @xmath928 is an isomorphism between @xmath929 and a line in @xmath872 , which contradicts to that @xmath929 is an elliptic curve . therefore ( [ spherical ] ) also holds for a line bundle @xmath52 . it is enough to show the @xmath931-stability of @xmath896 with respect to the triple ( [ triple:400 ] ) . by lemma [ lem : zg:4 ] , we have the following : @xmath932 ) \subset \left\ { \frac{1}{2 } \sqrt{\frac{d}{h^2 } } \times \mathbb{z}_{\ge 0 } \right\}. \end{aligned}\ ] ] we first consider the case @xmath933 . in this case , the imaginary part of @xmath934 is @xmath935 , that is the smallest positive value of the rhs of ( [ note : imm ] ) . therefore by lemma [ lem : cmin ] , the @xmath931-stability of @xmath896 follows from @xmath936 for any @xmath937 $ ] with @xmath938 . the above vanishing holds since @xmath939 is given by an iterated extensions of zero dimensional sheaves and objects @xmath940 $ ] for torsion free @xmath675-stable sheaves with @xmath941 . next we consider the case @xmath942 . in this case , @xmath943 , whose imaginary part is the twice of the smallest positive value @xmath944 of the rhs of ( [ note : imm ] ) . therefore , besides the vanishing ( [ vanish : p ] ) , we need to show the following : if there is an exact sequence in @xmath945 $ ] @xmath946 with @xmath947 , then we have the inequality in @xmath948 $ ] @xmath949 in order to show this , we first observe that @xmath950 $ ] is @xmath931-stable by ( * ? ? ? * proposition 7.4.1 ) . since @xmath896 is a subobject of @xmath622 in @xmath945 $ ] , and @xmath951 , we have @xmath952 since the image of @xmath931 is contained in @xmath953 , and @xmath954 , the above inequality implies that @xmath955 . it remains to exclude the following case : @xmath956 indeed , we see that there is no object @xmath937 $ ] such that @xmath957 . suppose that such an object @xmath132 exists , and let @xmath958 be the ( untwisted ) mukai vector of @xmath959 . then the condition @xmath960 is equivalent to @xmath961 this is a contradiction since both of @xmath962 and @xmath502 are integers . in this subsection , we study the case of @xmath74 . this case seems to be the most interesting case studied in this paper , as a construction of @xmath50 has to do with the study of coherent systems on the smooth projective curve @xmath5 . indeed in proposition [ prop : syst ] , we constructed an equivalence @xmath519 between the category of coherent systems on @xmath5 and the heart @xmath69 , given by @xmath963 below we abbreviate @xmath519 and regard any coherent system @xmath964 on @xmath5 as an object in @xmath69 . in the notation of subsection [ subsec : central2 ] , the polynomial @xmath653 is of type @xmath965 , which must belong to one of the classifications ( [ types : n=4 ] ) . there are two possibilities : @xmath966 the calabi - yau manifold @xmath639 is a k3 surface which is either a quartic surface or a double cover of @xmath872 , and it contains @xmath5 as an element in the linear system @xmath967 . the central charge @xmath40 is described in terms of coherent systems as follows : although @xmath69 is a noetherian abelian category , we have to take a little care since the condition in ( * ? ? ? * proposition 2.12 ) is not satisfied in this case . instead , we apply the argument used in ( * ? ? ? * theorem 2.29 ) . let @xmath974 be the subcategory consisting of objects @xmath975 such that @xmath502 is surjective . note that the right orthogonal complement @xmath976 consists of objects of the form @xmath977 . any object @xmath978 fits into the exact sequence @xmath979 showing that @xmath980 is a torsion pair on @xmath69 . furthermore , since @xmath972 on @xmath976 , we can easily see the following : an object @xmath336 is @xmath675-semistable if and only if for any exact sequence @xmath981 in @xmath333 , we have @xmath982 . * lemma 2.27 ) . ) since @xmath333 is a noetherian and artinian quasi - abelian category , an argument similar to ( * ? ? ? * theorem 2.29 ) shows that any object in @xmath333 admits a @xmath675-harder - narasimhan filtration . combined with ( [ ex : system ] ) , any object in @xmath69 admits a @xmath675-harder - narasimhan filtration . before discussing the construction of @xmath50 , we review clifford type theorem for coherent systems established by newstead - lange @xcite . for a smooth projective curve @xmath929 and @xmath983 , recall that the @xmath586-stability on @xmath984 is defined by the slope function @xmath985 here the above slope function is set to be @xmath888 if @xmath986 . _ ( ( * ? ? ? * theorem 2.1 , remark 2.3))_[thm : cli ] let @xmath929 be a smooth projective curve of genus @xmath987 , and @xmath983 . then for any @xmath586-stable coherent system @xmath988 with @xmath989 , we have @xmath990 moreover if @xmath929 is non - hyperelliptic , then the equality holds in ( [ ineq : cli ] ) only if @xmath964 is isomorphic to either @xmath991 by the @xmath675-stability , the morphism @xmath502 is surjective ( cf . remark [ rmk : surj ] ) hence @xmath994 . also the rhs of ( [ drc ] ) is @xmath79 if @xmath995 and @xmath996 if @xmath997 . in both cases , they are smaller than @xmath998 , so we may assume that @xmath999 . comparing the @xmath675-stability in ( [ slope : mu ] ) and the @xmath586-stability in ( [ slope : alpha ] ) , we see that the @xmath675-stable object @xmath964 is @xmath586-stable for @xmath1000 , since the set of quotient sheaves of @xmath76 with bounded above degrees is bounded . therefore , applying theorem [ thm : cli ] , we obtain the inequality @xmath1001 . it remains to check that the equality @xmath1002 does not hold if @xmath995 . in this case , @xmath5 is a quartic curve , so it is not hyperelliptic . also @xmath1003 is one dimensional , @xmath1004 is three dimensional , hence the objects ( [ exclude ] ) do not satisfy our assumption @xmath1005 . therefore the case @xmath1002 is excluded . as we discussed in subsection [ subsec : strategy ] , the slope function @xmath675 defines a torsion pair ( [ mu : tilting ] ) on @xmath69 , and the associated tilting @xmath70 is given by ( [ ag : tilting ] ) . we have the following result : obviously the triple ( [ triple:3 - 1 ] ) determines a stability condition if and only if the triple @xmath1007 determines a stability condition . by the construction of @xmath70 , any non - zero object @xmath1008 satisfies @xmath1009 . moreover lemma [ lem : zg:3 ] and lemma [ lem : crucial ] imply that if @xmath1010 , then @xmath1011 holds . therefore the triple ( [ triple:3 - 11 ] ) satisfies the condition ( [ htheta ] ) . also by lemma [ lem : zg:3 ] , the image of @xmath931 is discrete , which enables us to apply the same argument of ( * ? ? ? * proposition 7.1 ) to prove the harder - narasimhan property of the triple ( [ triple:3 - 11 ] ) . therefore the triple ( [ triple:3 - 11 ] ) , hence ( [ triple:3 - 1 ] ) , determines a stability condition . let @xmath1012 be the stability condition on @xmath3 determined by the triple ( [ triple:3 - 1 ] ) . we need to show that @xmath50 is gepner type with respect to @xmath39 . note that , in our situation , the triple ( [ triple:3 - 1 ] ) satisfies the inequality in ( [ condition : n=3 ] ) . therefore , by lemma [ lem : con : n=3 ] , it is enough to check the @xmath50-stability of @xmath704 , @xmath835 and @xmath1013 . the stabilities of these objects are proved in lemma [ lem : taupsi ] , lemma [ lem : check : c1 ] and lemma [ lem : tau2psi ] below . below we check the @xmath50-stability of the objects @xmath405 , @xmath835 and @xmath1013 . let @xmath1014 be the stability condition on @xmath3 determined by the triple ( [ triple:3 - 11 ] ) . it differs from @xmath50 by an action of @xmath78 , so it is enough to check the @xmath1014-stability of these objects . by lemma [ lem : tau : ox ] , the object @xmath405 is an object in @xmath69 , given by the coherent system @xmath1015 . it is @xmath675-stable with @xmath1016 , hence @xmath1017 \in { \mathcal{a}}_g$ ] . on the other hand , we have @xmath1018 is @xmath1019 , which is the smallest positive number in the rhs of ( [ , it is enough to check that there is no non - zero morphism from any object @xmath1020 with @xmath938 to @xmath1021 $ ] . since @xmath1022 \in { \mathcal{a}}_w[-1]$ ] , and @xmath1023 by sublemma [ im=0 ] below , there is no non - zero morphism from @xmath132 to @xmath1017 $ ] . [ im=0 ] a non - zero object @xmath1020 satisfies @xmath938 if and only if @xmath1023 , and it is given by an iterated extensions of @xmath675-stable coherent systems @xmath964 with @xmath1024 , and coherent systems of the form @xmath1025 for @xmath1026 . by corollary [ cor : canonical ] , we have the isomorphism ( again we abbreviate @xmath519 ) @xmath521 \cong ( { \mathcal{o}}_x \otimes r_1 \stackrel{s}{\to } { \mathcal{o}}_x(1))\end{aligned}\ ] ] such that @xmath502 is the natural evaluation map . in particular @xmath1027 is an isomorphism , and @xmath502 is surjective since @xmath604 is globally generated . then we apply sublemma [ sublem : line ] below to show that @xmath520 $ ] is @xmath675-stable . the slope @xmath1028)$ ] equals to @xmath1029 if @xmath995 and @xmath800 if @xmath997 . in the former case , @xmath1030 \in { \mathcal{a}}_g$ ] , and the imaginary part of @xmath1031)$ ] is @xmath618 , which is the smallest positive value of the rhs of ( [ -stability of @xmath1030 $ ] follows from the same argument of lemma [ lem : taupsi ] . in the latter case , @xmath520 \in { \mathcal{a}}_g$ ] and the imaginary part of @xmath1032)$ ] is @xmath800 . hence the @xmath1014-stability of @xmath520 $ ] follows from its @xmath675-stability and sublemma [ im=0 ] . suppose that @xmath502 is surjective and @xmath1027 is injective . let @xmath1036 be an exact sequence of coherent systems . it is enough to show the inequality @xmath1037 by our assumption , @xmath1038 and @xmath1039 , hence @xmath1040 is a zero dimensional sheaf . therefore the rhs of ( [ check : stable ] ) equals to @xmath944 , while the lhs of ( [ check : stable ] ) is less than @xmath944 . hence ( [ check : stable ] ) holds . applying @xmath36 to the exact sequence ( [ tau : ox ] ) , we obtain the distinguished triangle in @xmath3 @xmath1042 combined with lemma [ lem : tau : ox ] and corollary [ cor : canonical ] , the object @xmath1013 is obtained as a cone of the morphism of coherent systems : @xmath1043^{s } \ar[d]_{\gamma_1 } & { \mathcal{o}}_x(1 ) \ar[d]^{\gamma_2 } \\ { \mathcal{o}}_x \ar[r ] & { \mathcal{o}}_x . } \end{aligned}\ ] ] here @xmath502 is the evaluation map and @xmath1044 is the morphism of coherent systems . since @xmath1044 is non - zero , both of @xmath1045 , @xmath1046 are non - zero , hence they are surjective . therefore the object @xmath1047 $ ] is given by the coherent system @xmath1048 here @xmath1049 is the subspace of @xmath841 which vanishes at @xmath365 , and @xmath1050 is the ideal sheaf which defines @xmath365 . first suppose that @xmath995 . in this case , @xmath1049 is two dimensional , and @xmath1051 is surjective since any of two lines in @xmath872 determined by generic two elements in @xmath1049 intersect only at @xmath365 transversally . also @xmath1052 is injective since @xmath1027 is an isomorphism . therefore the coherent system ( [ tau:2 ] ) is @xmath675-stable by sublemma [ sublem : line ] . since @xmath1053 for the coherent system ( [ tau:2 ] ) , we have @xmath1047 \in { \mathcal{a}}_g$ ] , and it is @xmath50-stable . in particular @xmath1013 is an object in @xmath1054 . next suppose that @xmath997 . in this case , @xmath1049 is one dimensional , and @xmath1051 is not surjective at @xmath365 . in particular the coherent system ( [ tau:2 ] ) is not @xmath675-semistable . however we can show the @xmath931-stability , hence @xmath1014-stability , of ( [ tau:2 ] ) in the following way : the sheaf @xmath1055 is isomorphic to @xmath1056 for another point @xmath1057 , and ( [ tau:2 ] ) is isomorphic to the coherent system @xmath1058 where @xmath508 is a natural inclusion . there is an exact sequence of coherent systems @xmath1059 both of the objects @xmath1060 and @xmath1061 are @xmath675-stable with negative slopes , hence the object ( [ tau:2 ] ) is an object in @xmath70 . also the imaginary part of @xmath1062 is @xmath935 , which is the smallest positive value of the rhs of ( [ , it is enough to check that there is no non - zero morphism from @xmath1020 with @xmath938 to the object @xmath1063 . this follows from sublemma [ im=0 ] , since both of @xmath1060 and @xmath1061 are @xmath675-stable with negative slopes , and the exact sequence ( [ ex : system2 ] ) does not split . finally in this subsection , we study the case of @xmath1065 . in this case , @xmath5 is a finite number of smooth points , represented by points @xmath1066 for @xmath1067 . the heart @xmath69 is given by @xmath1068 by lemma [ lem : compute2 ] and lemma [ lem : equivalence ] , the heart @xmath69 is equivalent to the abelian category of representations of a certain quiver @xmath831 with relations . as in the previous subsection , we have the following possibilities : @xmath1069 { \bullet}+!d { } ( : ^{\pi^{(4)}}[ru]{\bullet}+!d { } , : _ { \pi^{(3)}}[rru]{\bullet}+!d { } , : _ { \pi^{(1)}}[rd]{\bullet}+!d { } , : ^{\pi^{(2)}}[rrd]{\bullet}+!d { } ) } \\ & ( a_1 , a_2 , d , \sharp x)=(3 , 1 , 6 , 2 ) , \quad { \mathcal{q}}= \xygraph { { \bullet}+d { } : ^{x_2}[rr ] { \bullet}+!d { } ( : ^{\pi^{(2)}}[ru]{\bullet}+!d { } , : _ { \pi^{(1)}}[rd]{\bullet}+!d { } ) } \end{aligned}\ ] ] also by corollary [ cor : quiver ] , in the @xmath995 case , we put the the following relation : @xmath1070 there is no relation in the @xmath997 case . by ( [ zdag : com ] ) and the above classification , the central charge @xmath40 is given as follows : [ zg:2 - 2 ] if we write the k - theory class of @xmath511 as @xmath1071= v_1 [ \mathbb{c}(1 ) ] + v_0 [ \mathbb{c}(0 ) ] + \sum_{j=1}^{\sharp x } w_j [ \psi({\mathcal{o}}_{p^{(j)}})]\end{aligned}\ ] ] for @xmath1072 , then we have @xmath1073 here we set @xmath1074 . by setting @xmath970 , we define the slope function @xmath675 on @xmath69 by @xmath1075 here we set @xmath1076 if @xmath1077 . the above slope function defines the @xmath675-stability on @xmath69 . since @xmath69 is noetherian and artinian , any object in @xmath69 admits a @xmath675-harder - narasimhan filtration . proposition 2.12 ) . ) we prepare the following lemma : in the case of @xmath997 , we have @xmath1080 , which is non - positive . if @xmath1081 , the condition @xmath1078 implies @xmath1082 . therefore @xmath52 decomposes into direct sums , which contradicts to the @xmath675-stability of @xmath52 . in the case of @xmath995 , the claim is equivalent to that if @xmath52 is @xmath675-stable with @xmath1083 , then @xmath1084 . let us represent @xmath52 as a representation of @xmath831 : @xmath1085 { \bullet}+!d{v^{(0 ) } } ( : ^{\pi^{(4)}}[ru]{\bullet}+!d{w^{(4 ) } } , : _ { \pi^{(3)}}[rru]{\bullet}+!d{w^{(3 ) } } , : _ { \pi^{(1)}}[rd]{\bullet}+!d{w^{(0 ) } } , : ^{\pi^{(2)}}[rrd]{\bullet}+!d{w^{(2 ) } } ) } \end{aligned}\ ] ] where @xmath1086 are @xmath1087-dimensional and @xmath846 are @xmath1088-dimensional . suppose by a contradiction that @xmath1089 . by the relation ( [ q : relation ] ) , we have the linear maps for @xmath1090 @xmath1091 because of the condition @xmath1092 , there is @xmath1090 such that @xmath1093 . also the @xmath675-stability of @xmath52 implies that @xmath1094 is surjective . therefore the assumption @xmath1089 implies that @xmath1095 , and there is a non - zero element @xmath1096 which is mapped to zero by ( [ map : v ] ) . let us consider the sub quiver representation of ( [ rep : q ] ) generated by @xmath1097 . by taking account the relation ( [ q : relation ] ) into consideration , the corresponding subobject @xmath1098 in @xmath69 has the k - theory class either @xmath1099 \ \mbox { or } \ [ \mathbb{c}(1 ) ] + [ \mathbb{c}(0 ) ] \ \mbox { or } \ [ \mathbb{c}(1 ) ] + [ \mathbb{c}(0 ) ] + [ \psi({\mathcal{o}}_{p^{(i ) } } ) ] . \end{aligned}\ ] ] hence we have @xmath1100 , which contradicts to the @xmath675-stability of @xmath52 with @xmath1078 . the triple ( [ triple:2 - 2 ] ) determines a stability condition if and only if the triple @xmath1102 determines a stability condition . by lemma [ zg:2 - 2 ] , lemma [ lem:2 - 2 ] and the construction of @xmath70 , the triple ( [ triple:2 - 20 ] ) satisfies the condition ( [ htheta ] ) . also since the image of @xmath931 is discrete , the same argument of ( ? ? ? * proposition 7.1 ) shows the harder - narashiman property of ( [ triple:2 - 20 ] ) . therefore the triples ( [ triple:2 - 2 ] ) , ( [ triple:2 - 20 ] ) determine stability conditions @xmath50 , @xmath1014 respectively . since they only differ by a @xmath78-action , by lemma [ lem : fur ] , it is enough to check the @xmath1014-stability of @xmath704 , @xmath835 and @xmath1103 . these are checked in lemma [ check:1 ] , lemma [ check:2 ] and lemma [ check:3 ] below . by lemma [ lem : tau : ox ] , we have @xmath1104 . by the exact sequence ( [ tau : ox ] ) , the object @xmath704 is @xmath675-stable with non - positive slope . therefore @xmath704 is an object in @xmath70 . in the @xmath995 case , we have @xmath1105 . since the image of @xmath931 is @xmath1106 , this immediately implies that @xmath405 is @xmath1014-stable . in the @xmath997 case , note that @xmath1107 on @xmath70 is contained in @xmath1108 . since @xmath1109 , by lemma [ lem : cmin ] , it is enough to check that @xmath1110 for any @xmath1020 with @xmath1111 . by our construction of @xmath70 , we have @xmath1023 , and it is @xmath675-semistable with @xmath1112 . since @xmath704 is @xmath675-stable with negative slope , we have @xmath1110 . since @xmath835 is @xmath675-stable with @xmath1113 , we have @xmath520 \in { \mathcal{a}}_g$ ] . in the @xmath995 case , note that @xmath1107 is contained in @xmath1114 . since @xmath1115)=-1 $ ] , by lemma [ lem : cmin ] , it is enough to check that @xmath1116)=0 $ ] for any @xmath1020 with @xmath1111 , which follows since @xmath132 is an object in @xmath69 . in the @xmath997 case , let @xmath52 be a non - trivial quotient of @xmath520 $ ] in @xmath70 . since @xmath1032)=-1 $ ] , it is enough to check that @xmath1079 . to check this , note that we have @xmath726 since @xmath835 is a simple object in @xmath69 . let us write the k - theory class of @xmath52 as in ( [ k - th ] ) . since @xmath1117 is non - positive , it is enough to exclude the case of @xmath1081 . if @xmath1081 , then @xmath52 is a direct sum of objects @xmath835 and @xmath366 for @xmath367 . then there is a non - zero morphism from @xmath520 $ ] to @xmath366 , which contradicts to lemma [ lem : compute2 ] . we first discuss the case of @xmath995 . in this case , by corollary [ cor : quiver2 ] , the object @xmath1118 $ ] is an object in @xmath69 represented by the following quiver representation @xmath1119 { \bullet}+!d{r_2 } ( : ^{\pi^{(4)}}[ru]{\bullet}+!d{\mathbb{c } } , : _ { \pi^{(3)}}[rru]{\bullet}+!d{\mathbb{c } } , : _ { \pi^{(1)}}[rd]{\bullet}+!d{\mathbb{c } } , : ^{\pi^{(2)}}[rrd]{\bullet}+!d{\mathbb{c } } ) } \end{aligned}\ ] ] here @xmath841 is two dimensional and @xmath1120 is three dimensional . we first show that @xmath1118 $ ] is @xmath675-stable . let @xmath52 be a subobject of @xmath1118 $ ] in @xmath69 given by a representation ( [ rep : q ] ) , with @xmath1121 and @xmath1122 . let @xmath522 be the subset of @xmath1123 such that @xmath850 . similarly to the proof of lemma [ lem:30:check ] , we have the inequality @xmath1124 . also we may assume that @xmath1125 , since otherwise @xmath76 coincides with @xmath1118 $ ] . therefore we obtain the inequality @xmath1126 which implies that @xmath1127)=1/4 $ ] . therefore @xmath1118 $ ] is @xmath675-stable with positive slope , hence we have @xmath1128 \in { \mathcal{a}}_g$ ] . since @xmath1129 ) = -1 + \sqrt{-1}$ ] , and the image of @xmath931 is @xmath1130 , by lemma [ lem : cmin ] , the @xmath1014-stability of @xmath1128 $ ] follows if we check that @xmath1131)=0 $ ] for any @xmath1020 with @xmath1111 . this follows since @xmath1132 and @xmath1128 \in { \mathcal{a}}_w[-1]$ ] . next we discuss the case of @xmath997 . by lemma [ lem : filt : a ] , the object @xmath1118 \in { \mathcal{a}}_w$ ] is represented by the following quiver representation @xmath1133 { \bullet}+!d{\mathbb{c } } ( : ^{{\textrm{id}}}[ru]{\bullet}+!d{\mathbb{c } } , : _ { { \textrm{id}}}[rd]{\bullet}+!d{\mathbb{c } } ) } \end{aligned}\ ] ] the above object is obviously @xmath675-stable . since @xmath1134)=-3/4 $ ] , it follows that @xmath1118 \in { \mathcal{a}}_g$ ] . also we have @xmath1135)=-(1/2 + 3\sqrt{-3}/2)$ ] , and @xmath1107 is contained in @xmath1108 on @xmath70 . applying lemma [ lem : cmin ] , the @xmath1014-stability of @xmath1118 $ ] follows if we check that @xmath1136)=0 $ ] for any @xmath1020 with @xmath1111 . since such @xmath132 is a @xmath675-semistable object in @xmath69 with @xmath1112 , and @xmath1118 $ ] is @xmath675-stable with negative slope , it follows that we have @xmath1136)=0 $ ] . y. toda . generating functions of stable pair invariants via wall - crossings in derived categories . , 59:389434 , 2010 . new developments in algebraic geometry , integrable systems and mirror symmetry ( rims , kyoto , 2008 ) .	we introduce the notion of gepner type bridgeland stability conditions on triangulated categories , which depends on a choice of an autoequivalence and a complex number . we conjecture the existence of gepner type stability conditions on the triangulated categories of graded matrix factorizations of weighted homogeneous polynomials . such a stability condition may give a natural stability condition for landau - ginzburg b - branes , and correspond to the gepner point of the stringy k@xmath0hler moduli space of a quintic 3-fold . the main result is to show our conjecture when the variety defined by the weighted homogeneous polynomial is a complete intersection of hyperplanes in a calabi - yau manifold with dimension less than or equal to two .
in this appendix we give the explicit derivation of the equations of motion ( [ eqn : psol1]-[eqn : psol4 ] ) , the laplace transformed equations of motion ( [ eqn : rsol1],[eqn : rsol2 ] ) , and the bosonic correlation function ( [ eqn : bosoncf ] ) . we note that parts of these derivations can be found in similar form in the literature ( cf . @xcite ) , but we here give an alternate and full account in consistent notation for the benefit of the reader . at this point we make the born approximation ( assuming the density matrix of system and bath are factorable ) @xmath98 the bosonic correlation function is defined as @xmath99 : @xmath100.\ ] ] where the subscript @xmath101 represents the bosonic degrees of freedom . we substitute this into ( [ eqn : efmo ] ) and by assuming there is no initial coherence in the system we obtain ( [ eqn : psol1 ] - [ eqn : psol4 ] ) . the bosonic correlation function ( [ eqn : bosoncf ] ) for an oscillator with a single mode in a thermal state is defined as : @xmath102 , \qquad \rho_b = \frac{\exp(- \beta \omega { \ensuremath{{\ensuremath{a^\dagger}}{\ensuremath{a}}}})}{\text{tr}_b [ \exp(- \beta \omega { \ensuremath{{\ensuremath{a^\dagger}}{\ensuremath{a } } } } ) } = \frac{1}{z } \exp(- \beta \omega { \ensuremath{{\ensuremath{a^\dagger}}{\ensuremath{a } } } } ) . \label{eqn : bcf2}\ ] ] this can be evaluated in different ways , one of which is presented below . starting from the time dependence of the displacement operator in the interaction picture : @xmath103 or , alternatively , through the time dependence of creation and annihilation operators : @xmath104 in order to perform the trace @xmath105 in the number state basis , we need to know the action of @xmath106 and @xmath107 on a number state @xmath108 . the first simply evaluates to @xmath109 and the latter gives the so - called displaced number state @xmath110 . the displaced number state can be expanded in the number state basis @xmath111 with ( see , e.g. , oliviera et al @xcite or m. crisp @xcite ) @xmath112 where @xmath113 is an associated laguerre polynomial . this is only valid for @xmath114 , but for @xmath115 the displacement operator , or rather its hermitian conjugate , can be made to act on @xmath116 instead of on @xmath108 . we use eq . ( [ eqn : dotd ] ) for the displacement operator and the property @xmath117 d(x + y)$ ] to evaluate eq . ( [ eqn : bcf2 ] ) . this leads to a series of the following form @xmath118 } \sum_{n = 0}^{\infty } e^{-\beta \omega n } l_n[2 { { \left\vert\alpha\right\vert}}^2 ( 1 - \cos[\omega ( t - t ' ) ] ) ] . \label{eqn : bcf - single - series}\ ] ] by virtue of the property @xmath119 $ ] and with @xmath120 and @xmath121 we finally arrive at the analytical result : @xmath122 \exp[-2{{\left\vert\alpha\right\vert}}^2(1-\cos[\omega ( t - t')])(n+1/2 ) ] . \label{eqn : bcf - analytical}\ ] ] note that this expression agrees with mahan s result for a single mode ( ref @xcite , section 4.3 ) @xmath123 . mahan derives this in a similar fashion but without using eqs . ( [ eqn : dnstate ] , [ eqn : dncoeffs ] ) . instead , he uses the ` feynman disentanglement of operators ' to arrive at an equivalent infinite series of laguerre polynomials .	a quantum two level system coupled to a harmonic oscillator represents a ubiquitous physical system . new experiments in circuit qed and nano - electromechanical systems ( nems ) achieve unprecedented coupling strength at large detuning between qubit and oscillator , thus requiring a theoretical treatment beyond the jaynes cummings model . here we present a new method for describing the qubit dynamics in this regime , based on an oscillator correlation function expansion of a non - markovian master equation in the polaron frame . our technique yields a new numerical method as well as a succinct approximate expression for the qubit dynamics . we obtain a new expression for the ac stark shift and show that this enables practical and precise qubit thermometry of an oscillator . the qubit - oscillator model has gone by many names in many fields , owing its tenacity to the breadth of its applicability : it is the simplest non - trivial model of the interaction between light and matter . at its inception it was used to describe the interaction of an atom with a magnetic field @xcite , and referred to thereafter as the rabi model . in the subsequent decades it has been extensively studied in quantum optics @xcite and cavity qed @xcite . physical chemists have used a ` vibration - dimer ' model to study the spectra of molecules @xcite . applying the rotating wave approximation ( rwa ) to the rabi model yields the jaynes cummings model ( jcm ) @xcite , which is valid when the detuning between the qubit transition frequency @xmath0 and the resonator frequency @xmath1 is negligible @xmath2 and the coupling between the qubit and oscillator is weak @xmath3 @xcite . this is an excellent approximation in the case of cavity qed where typical coupling strengths are of order @xmath4 . the jcm can be extended to incorporate tunnelling , and has provided an adequate description of experiments for decades , but a new era of experiments are pushing beyond its boundaries in terms of both coupling and detuning . circuit qed experiments couple superconducting qubits to lc and waveguide resonators , allowing coupling strengths up to @xmath5 , recently enabling demonstrations of the breakdown of the jcm @xcite . superconducting qubits coupled to nanomechanical resonators ( nr ) generally have more modest coupling strengths @xcite , but combined with large detuning they could also operate outside the validity of the jcm @xcite . the hamiltonian for the rabi model can be decomposed into three parts : @xmath6 the qubit , atom or two level system is described by : @xmath7 where @xmath8 and @xmath9 are the pauli spin operators . they describe a two level system with an energy splitting @xmath10 and a spontaneous tunnelling between the states at a rate @xmath11 . in isolation such a system would undergo rabi oscillations with a frequency @xmath12 . the hamiltonian of the oscillator is : @xmath13 where @xmath1 is the frequency of the oscillator and @xmath14 and @xmath15 are its creation and annihilation operators respectively . note we have neglected the zero point energy . the hamiltonian for the interaction between the two is : @xmath16 where @xmath17 is the coupling strength between the qubit and oscillator . recent experimental progress has sparked a renewed theoretical interest in extending solutions of ( [ eqn : hambreak ] ) beyond the rwa . for instance , a change of basis prior to applying the rwa leads to a generalised rwa that should be valid outside the very weak coupling limit @xcite . however , this is limited to the case of @xmath18 . as an alternative approach , van vleck perturbation theory @xcite has been used to investigate the dynamics in the ultra strong ( @xmath19 ) coupling regime @xcite . this approach contains the splitting and tunnelling elements , but it is perturbative in the latter and fails to recover the jcm in the weak coupling limit . this approach is therefore more applicable to circuit qed , rather than the more modest couplings achieved in cooper pair box ( cpb ) coupled to nr systems . an analytic expression for the eigenspectrum of the full rabi model was very recently found by braak @xcite , a surprising and significant result for such a long standing problem . however , it is too early to tell how much this solution can reveal about the physical properties and dynamics of the system . in addition to solving the model braak proved that it is non - integrable i.e. the time - dependence of important properties can not be found in closed form . there is therefore still a need for approximate results governing areas of particular experimental interest . in order to simplify the expression and extend the validity of the approximations that we will subsequently describe into the strong coupling regime , we first perform a ` polaron ' transformation @xcite . this unitary hamiltonian transformation ( @xmath20 ) is equivalent to dressing qubit excitations with the vibrational modes to form quasi - particles called polarons . with @xmath21 and @xmath22 we obtain @xmath23 where @xmath24 is the displacement operator . we have neglected a term proportional to the identity @xmath25 , which does not influence the dynamics . the first two terms involve the qubit and oscillator individually and so can be removed by going to the interaction picture . we insert the resulting hamiltonian into the von neumann equation and then derive equations of motion for the qubit ( see appendix and refs . @xcite ) : @xmath26 where @xmath27 and @xmath28 are the time dependant population elements of the qubit s reduced density matrix . the coherences are given by : @xmath29 , \label{eqn : psol3 } \\ \rho_{10}(t ) & = -i\frac{\delta}{2 } \int_0^t \mathrm{d}t ' e^{i \epsilon\tau } \left [ \rho_{00}(t ' ) c(\tau)-\rho_{11}(t ' ) c^(\tau ) \right ] , \label{eqn : psol4}\end{aligned}\ ] ] where @xmath30 and @xmath31 and @xmath32 are the correlation function of the oscillator and its complex conjugate respectively . in deriving these equations , we have employed the born approximation , i.e. we have assumed that the vibrational mode and the qubit states can be factored at all times . physically , this corresponds to an oscillator that thermalises on a timescale faster than that characteristic of the qubit dynamics . the equations of motion take the form of a system of integro - differential equations involving the bosonic correlation function and its complex conjugate . laplace transforming the equations of motion yields a set of simultaneous equations that can be solved algebraically : @xmath33 } { s^{2}+ s \left ( \frac{\delta}{2 } \right)^{2 } \left[c'_{-}+c''_{-}+c'_{+}+c''_{+ } \right]}\label{eqn : rsol1}\\ & r_{10}(s ) & = -i \frac{\delta}{2 } \left [ ( c'_{- } + c''_{-})r_{00}(s ) -\frac{1}{s}c''_{- } \right ] \label{eqn : rsol2}\end{aligned}\ ] ] where @xmath34 is our laplace space variable , @xmath35 and @xmath36 are the laplace transforms of @xmath37 and @xmath38 , @xmath39 is the initial population of the ground state and @xmath40 and @xmath41 are the laplace transforms of the correlation function and its conjugate respectively . it is sufficient to solve these two equations alone because from their solutions the behaviour of the other density matrix elements can be trivially derived . to obtain expressions for the dynamics of eqns ( [ eqn : rsol1 ] ) and ( [ eqn : rsol2 ] ) in the time domain we need to find the laplace transform of the bosonic correlation and its conjugate , solve and then take the inverse laplace transform of the equations . the correlation function is defined as @xmath42\ ] ] which evaluates to ( see appendix ) : @xmath43 unfortunately , it is not straightforward to laplace transform this expression directly , so we employ the jacobi - anger series expansion : @xmath44 where @xmath45 is an arbitrary complex number and @xmath46 is the modified bessel function of order @xmath47 and argument @xmath45 . by exploiting an angle addition identity we can rewrite ( [ eqn : mah ] ) as : @xmath48 where @xmath49 and @xmath50 . using eqn ( [ ja ] ) this gives : @xmath51 where @xmath52 is the average oscillator occupation number . in this form the correlation function can be laplace transformed trivially . the bessel function weighting of the series means it converges very rapidly , in fact only a few terms of the series need to be retained to accurately capture the dynamics . for experimentally relevant parameters ( i.e. low temperatures and moderate to strong coupling ) only a single term dominates . the corresponds to the regime where : @xmath53 comparison of the single term approximation ( red ) and a numerically exact approach ( blue ) for different coupling strengths . uncoupled rabi oscillations are also shown as a reference ( green ) . left : the population @xmath37 in the time - domain . right : the same data in the frequency domain . the full numerical solution was fourier transformed using matlab s fft algorithm . other parameters are @xmath54 ghz , @xmath55 ghz , @xmath56 mhz and @xmath57 mk . ] physically this corresponds to retaining only interactions that conserve the total phonon number in the oscillator , thus complementing the underlying born approximation , which assumes the oscillator remains in thermal equilibrium . including only the dominant zeroth term in the series allows the equations ( [ eqn : rsol1 ] ) and ( [ eqn : rsol2 ] ) to be inverse laplace transformed : @xmath58 @xmath59 @xmath60 where @xmath61 . from eqn ( [ eqn : rho3 ] ) we can see that the presence of the oscillator alters the tunnelling rate by a factor @xmath62 , essentially corresponding to a temperature dependent ac stark shift on the qubit due to the presence of the single oscillator mode @xcite . in contrast to previous work our expression is not confined to the weak coupling or large detunning limit , but still takes a surprisingly simple closed form . figure [ fig:1 ] shows a comparison of the dynamics predicted using these expressions and a numerically exact approach . the latter are obtained by imposing a truncation of the oscillator hilbert space at a point where the dynamics have converged and any higher modes have an extremely low occupation probability . our zeroth order approximation proves to be unexpectedly powerful , giving accurate dynamics well into the strong coupling regime ( @xmath63 ) and even beyond this it still captures the dominant oscillatory behaviour , see figure [ fig:1 ] . stronger coupling increases the numerical weight of higher frequency terms in the series , causing a modulation of the dynamics . the approximation starts to break down at ( @xmath64 ) . the equations ( [ eqn : rho0 ] ) and ( [ eqn : rho1 ] ) are obviously unable to capture the higher frequency modulations to the dynamics or any potential long time phenomena like collapse and revival , but these are unlikely to be resolvable in experiments in any case . nonetheless , it is worth pointing out that even in this strong coupling case the base frequency of the qubit dynamics is still adequately captured by our single term approximation . our methodology can be used to predict dynamics of nanomechanical resonators connected to either quantum dots or superconducting qubits . the criterion for the single term approximation to be valid is readily met by current experiments such as those presented in refs . @xcite and their parameters yield near perfect agreement between numerical and analytic results . most experiments operate in a regime where the qubit dynamics are not greatly perturbed by the presence of the oscillator , which has a much lower frequency ( @xmath65 ghz , @xmath66 ghz ) . in figure [ fig:1 ] , we chose @xmath67 mhz , because this better demonstrates the effect of the oscillator on the qubit . these parameters can be achieved experimentally using the same qubit design but with an oscillating voltage applied to the cpb bias gate @xcite . however , we stress the accuracy of our method is not restricted to this regime . main panel : comparison of dynamics calculated from truncating ( [ sin5 ] ) at @xmath68 ( red ) and a numerically exact approach ( blue ) . lower left : fourier transform of the dynamics . lower right : the numerical weight of the @xmath69 term in the series expansion of ( [ sin5 ] ) , showing there are still only two dominant frequencies at @xmath70 and @xmath71 . parameters : @xmath72 , @xmath55 , @xmath73 , @xmath74 , @xmath75 , @xmath76 and @xmath77 . ] including extra terms in the series expansion ( [ sin5 ] ) makes the time dependence of the qubit dynamics analytically unwieldy , because the rational function form of the series leads to a complex interdepence of the positions of the poles in ( [ eqn : rsol1 ] ) . however , if the values of the parameters are known the series can truncated at ( @xmath78 ) to give an efficient numerical method to obtain more accurate dynamics , extending the applicability of our approach beyond the regime described by ( [ eqn : crit ] ) . this is demonstrated in fig . [ fig:2 ] , where the dynamics are clearly dominated by two frequencies an effect that could obviously never be captured by a single term approximation . there is a qualitative agreement between the many terms expansion and full numerical solution , particularly at short times . we would not expect a perfect agreement in this case because the simulations are of the dynamics in the large tunnelling regime ( @xmath79 ) , and the polaron transform makes the master equation perturbative in this parameter . nonetheless , the rapid convergence of the series is shown in fig . [ fig:2 ] ; @xmath80 is sufficient to calculate @xmath37 and @xmath38 with an accuracy only limited by the underlying born approximation . the asymmetry of the amplitudes of the terms in the series expansion of ( [ sin5 ] ) is due to the exponential functions in the series . demonstration of qubit thermometry : @xmath81 is the temperature supplied to the numerical simulation of the system and @xmath82 is the temperature that would be predicted by fitting oscillations with frequency ( [ eqn : rho3 ] ) to it . the blue line is the data and red line shows the effect of a 10khz error in the frequency measurement ; the grey dashed line serves as a guide to the eye . the lower inset shows the variation of the qubit frequency @xmath0 with temperature . the upper inset shows the dependence of the absolute error in the prediction against the signal length ( see text ) . other parameters are : @xmath54 ghz , @xmath83 ghz , @xmath84 , @xmath85 mhz ] we now discuss the application of our model system to the measurement of the temperature of an oscillator by observing the coupled qubit . we picture a situation in which the qubit oscillation frequency @xmath0 is the measured quantity . the tunnelling , coupling strength and energy splitting are usually within the control of the experimentalist ( or are at least known constants ) , and this yields the possibility of using the measured @xmath0 to estimate the temperature . a related idea was recently used in the calibration of a seminal resonator experiment @xcite to verify that the oscillator was in its ground state ( a critical part of the work ) . in that case , the authors used a comparison of numerical results for different occupation numbers @xmath86 with the measured population in the excited state of the qubit after a certain interaction time . a theoretical study of the same approach was preformed in @xcite , where the system was described by the jcm without a tunnelling term . in contrast , we here propose that simple analytic expressions for the qubit dynamics that are valid beyond the weak coupling regime , like our own , can be used to directly measure the temperature and hence @xmath86 of the oscillator , simply by observing the effective qubit rabi frequency @xmath0 . figure [ fig:3 ] demonstrates this idea , showing that by measuring @xmath0 and fitting it to our expression ( [ eqn : rho3 ] ) , we can obtain submilli - kelvin precision in the experimentally relevant regime of 20 - 55 mk . at low temperatures the single term frequency plateaus , causing the accuracy to break down . in the higher temperature limit , we also see a deviation from the diagonal , this is to be expected as we leave the regime of validity described by ( [ eqn : crit ] ) . naturally accuracy in this region could be improved by retaining higher order terms in ( [ sin5 ] ) , but this would become a more numeric than analytic approach . the upper inset shows the dependence of the accuracy of the prediction on the number of points ( at a separation of 1ns ) sampled from the dynamics . the accuracy increases initially as more points improve the fitted value of @xmath0 , however after a certain length the accuracy is diminished by long term envelope effects in the dynamics not captured by the single term approximation . we note that the corresponding analysis in the frequency domain would not be equally affected by the long time envelope , however a large number of points in the fft is then required in order to obtain the desired accuracy . the lower inset of figure [ fig:3 ] shows the direct dependence of @xmath0 on the temperature . the temperature range with steepest gradient and hence greatest frequency dependence on temperature varies with the coupling strength ; thus the device could be specifically designed to have a maximal sensitivity in the temperature range of the most interest . in summary , we have developed and explored a new approach to the qubit - oscillator model , finding succinct expressions for the qubit dynamics . in contrast to previous theoretical approaches , our expressions are valid in the stronger coupling regime that is rapidly gaining experimental relevance . we have further proposed an application of our model enabling precise temperature measurements of the oscillator mode . this could be used either as part of the calibration of an oscillator experiment or as a tuned , standalone device . we thank nikesh dattani , tom stace , gerard milburn , and ahsan nazir for fruitful discussions . this work was supported by the epsrc , the national research foundation and ministry of education , singapore , and the royal society . 25 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) , issn . , * * , ( ) , issn . , _ _ ( , ) . , , , , , , , , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , , , * * , ( ) . , , , , , , , , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , _ _ ( , ) , ed . , _ _ ( , ) , ed . , isbn . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , , , , * * , ( ) . , , , ( ) , . , , , , , * * , ( ) . , * * , ( ) .
in the standard electroweak theory , flavor - changing neutral current ( fcnc ) processes are strongly suppressed and can only occur via higher - order effects . hence , these processes will be sensitive to unknown particles or interactions that contribute in higher - order loop diagrams . such processes are ideal places to look for signals of new physics beyond the standard model ( sm ) . in the sm , the fcnc @xmath0 decay is predominantly a cp conserving process . its branching fraction is sensitive to the real part of the @xmath10 transition amplitude , while the related decays @xmath11 and @xmath12 sense the imaginary part and absolute value , respectively . like these decays , @xmath0 is theoretically clean and uncertainties in the hadronic matrix element can be removed by using the measured branching fraction of its relevant semileptonic decay @xmath13 . the sm predicts the branching fraction to be @xmath14 @xcite . although the prediction is solid in the sm , there is a possibility of enhancements from new physics contributions . in fact , phenomenological analyses give constraints on possible enhancements by up to an order of magnitude within the allowed range of new physics parameters from known kaon decays , including the measured branching fraction of @xmath12 @xcite . in addition , a new particle @xmath8 which decays into invisible particles , can also contribute to the three body decay @xmath15 . there is also a possibility of having new pseudoscalar particles , as predicted by several supersymmetry models . searches for the @xmath16 or @xmath17 modes and comparisons with the models have been made @xcite , and similar test should also be tried in the @xmath18 final state . the e391a experiment at the kek 12-gev proton synchrotron made the first search for the @xmath0 decay , based on the data sample taken in the first stage ( run - i ) . an upper limit on the branching fraction of 4.7@xmath7 at the 90% confidence level ( c.l . ) for @xmath19 was obtained @xcite . unfortunately the run - i data were compromised by large neutron backgrounds from a vacuum membrane that hung in the beam . this article is based on the e391a data taken in the periods from february to april in 2005 ( run - ii ) and from november to december in 2005 ( run - iii ) , after correcting the problem . these data had six times more @xmath2 decays and were analyzed with improved methods . the e391a detector was primarily designed to search for the @xmath20 decay @xcite . it had a csi calorimeter to detect two photons from a @xmath21 decay and hermetic veto counters to ensure that no other visible particles existed in the final state ( fig . [ fig : detector ] ) . to avoid backgrounds from interactions of the beam particles with air , most of the detector components were placed in a vacuum vessel . the csi calorimeter consisted of 496 blocks of @xmath22 @xmath23 undoped csi crystal . a @xmath24 @xmath25 beam hole was located at the center of the calorimeter to allow the beam particles to pass through . the main barrel ( mb ) and front barrel ( fb ) counters were sandwiches of lead plates and plastic - scintillation counters with 13.5 @xmath26 and 17.5 @xmath26 , respectively , and formed cylindrical walls surrounding the @xmath27 decay volume . collar shaped counters ( cc00 , cc02@xmath2807 ) were placed around the beam line for vetoing photons along the beam axis . charged particles that would hit the csi calorimeter were identified and rejected by energy deposits in a charge veto ( cv ) scintillation - counter hodoscope , located 50 cm upstream ; it covered the front face as well as the outer wall of the beam - hole area . beam - hole charge veto ( bhcv ) and back - anti ( ba ) counters were located in the downstream region along the beam axis and outside of the vacuum vessel . the bhcv consisted of eight 3-mm - thick plastic - scintillation counters , which were arranged to fully cover the downstream area of the beam hole . the ba was made of six superlayers , each consisting of a lead / scintillator sandwich and quartz blocks for run - ii . for run - iii the ba had five superlayers , where the lead / scintillator sandwich part was replaced by the lead tungsten oxide ( pwo ) crystals ; this change was the only difference in the e391a detector system between the run - ii and run - iii periods . data acquisition was made with a hardware trigger that required two or more electromagnetic showers in the csi calorimeter with total energy @xmath29 60 mev , and no obvious activity in the cv and other veto counters . loose on - line veto requirements were imposed , which ensured flexibility of setting energy thresholds consistently for the real data and the detector simulation during the off - line analysis . the @xmath30 decay should produce exactly four showers in the csi calorimeter and nothing else in all the other counters . these hits should also satisfy an in - time requirement to suppress backgrounds . the experimental signature of @xmath0 is four photons in the final states with a non - zero missing mass and a transverse momentum . photons were registered in the csi calorimeter as energy deposits in adjacent blocks ( clusters ) . the most probable hit position of the photon was obtained from the shower energy distribution among the csi blocks . events with four photon clusters in the csi were selected for further off - line processing . reconstruction of a pair of photons was made by assuming that they were from a @xmath21 that decayed on the beam axis ( @xmath31-axis ) . multiple pairings of four photons to reconstruct two @xmath21s were eliminated by taking the combination that gave the minimum @xmath32 value in the difference in @xmath31 vertices . a cut on the minimum @xmath32 for the common @xmath31 vertex @xmath33 was effective for good @xmath21 identification . also , requiring a large difference in the lowest two @xmath32 values was effective in reducing the combinatorial error . we required that the common @xmath21-decay vertex point @xmath33 should be in the decay volume 300 cm @xmath34 500 cm . the veto conditions on the mb , cv , and collar counters were carefully tuned . among these counters , the mb played a major role in vetoing photons undetected by the csi calorimeter . if too low an energy - threshold was imposed for all of the mb counters , back - splash from the csi / cv surface and electromagnetic shower leakage from the outer part of the csi calorimeter to the downstream area of the mb would cause acceptance loss . thus , we applied a tighter threshold to the upstream region of the mb to detect photons with a high efficiency , and a looser threshold to the downstream region of the mb to keep a large acceptance for the signal @xcite . suppression of the fusion cluster ( misidentification of two close - by clusters as one cluster ) was made by using the neural network ( nn ) technique trained by the stand - alone photons and the fusion clusters from the @xmath35 monte carlo ( mc ) simulation sample . further suppression was made on the shower shape of the cluster , and by examining the energy distribution and its rms fraction @xcite . [ h ] vs. effective mass ( @xmath36 ) plots of the @xmath37 system from run - iii mc simulations for ( a ) @xmath19 , ( b ) @xmath38 , ( c ) @xmath35 , and ( d ) run - iii real data . the statistics of the mc events in ( b ) and ( c ) are normalized to those of the real data . the signal region is defined to be inside of the triangular area in the plots . a curved line in ( a ) indicates the kinematical limit for the @xmath19 decay . , title="fig:",height=321 ] a geant3-based @xcite mc simulation was performed to define the signal region for @xmath0 and to estimate the acceptance . the @xmath0 decay was generated with the v - a interaction , and the events were processed with the same reconstruction code as the real data . among the incoming @xmath2 particles to the detector , 2.66@xmath4 decay within the fiducial decay volume . a total of @xmath39 incoming @xmath2 events were generated for both run - ii and run - iii . the generated events were overlaid with accidental hits on the counters accumulated by special triggers during the data taking . after applying the selection criteria described above , a two - dimensional distribution of the @xmath37 effective mass @xmath36 and the transverse momentum @xmath40 for the accepted events was examined . the result of the simulation for @xmath41 in the run - iii condition is shown in fig . [ fig : mcdata](a ) . the signal region was defined as a right triangular area shown in fig . [ fig : mcdata ] . the vertical edge of the triangle is located at the lower bound of the @xmath37 effective mass ( 0.268 gev/@xmath9 ) , and spanned the @xmath40 values between 0.1 to 0.2 gev/@xmath42 . the horizontal edge defined the lower bound of @xmath40 ( 0.1 gev/@xmath42 ) to avoid contamination from the @xmath43 decay ; it spanned the @xmath36 values between 0.268 to 0.450 gev/@xmath9 . the detector acceptance was calculated by taking the ratio of the number of events in the signal region to the number of @xmath2 particles that decayed in the fiducial vertex region . a list of the reduction factors that contributed to the acceptance after the @xmath2s decayed in the fiducial volume are summarized in table [ tab : reduction ] . the acceptances obtained for run - ii and run - iii were consistent , and the result was @xmath44 for each run period . lcl cuts & reduction & resultant + & factor & acceptance + @xmath27 decay in the fiducial volume & & 1.00 + + hardware trigger & four - cluster- & & + candidate in the csi calorimeter & 0.0499 & @xmath45 + + tighter csi fiducial & 0.616 & @xmath46 + + rejection of fused cluster & 0.361 & @xmath47 + + req . good cluster & 0.295 & @xmath48 + + @xmath32 cut for two @xmath21 vertex matching & 0.606 & @xmath49 + + other misc . requirements on csi & 0.829 & @xmath50 + + mb and upstream vetoes & 0.621 & @xmath51 + + downstream vetoes & 0.853 & @xmath52 + + req . signal box ( after all the cuts ) & 0.346 & @xmath53 + to study the dominant backgrounds , mc events of the @xmath54 decay with the statistics equivalent to 11.2 and 11.9 times the run - ii and run - iii data , and the @xmath43 decay equivalent to 0.80 and 2.52 times the run - ii and run - iii data , respectively , were generated with the same detector conditions as the real data . within these statistics , no mc events of these processes were found in the @xmath55 signal region for both the run - ii and run - iii conditions . the mc results of the @xmath54 and @xmath43 decays for the run - iii condition , with the same statistics as the real data , are shown in figs . [ fig : mcdata](b ) and ( c ) , respectively , together with the run - iii real data ( fig . [ fig : mcdata](d ) ) . the @xmath54 plot of @xmath40 vs. @xmath36 showed clear clustering well removed from the signal region of @xmath0 ( fig . [ fig : mcdata](b ) ) . events from the @xmath43 decay were spread over the @xmath40 vs. @xmath36 plot without having clear boundaries ( fig . [ fig : mcdata](c ) ) . the data plot ( fig . [ fig : mcdata](d ) ) outside of the masked signal region was characterized by the sum of fig . [ fig : mcdata](b ) and ( c ) , except for a small number of events that extended to higher @xmath40 from the @xmath56 cluster region . these latter events are considered to be due to @xmath57 regeneration in the detector , and are not simulated in the mc . they will be discussed in sec . the mc study showed that most events stayed close to the signal region except that those for @xmath58 were from @xmath54 and @xmath43 decays . the main background source was @xmath43 . the number of @xmath27 particles that decayed in the fiducial vertex region was estimated by detecting @xmath59 with the same @xmath21 selection criteria in the csi calorimeter as for @xmath60 . comparison with the mc simulation of @xmath61 , with the same @xmath21 selection criteria as the real data , along with overlaid accidental events , gave the number of @xmath27 decays as @xmath62 for run - ii and @xmath63 for run - iii , respectively . to estimate the systematic uncertainties , the fractional difference between the data and the mc simulation in each selection criteria was examined , and the quadratic sum weighted by the effectiveness of each of the acceptance determinations @xcite was taken . the calculated values were 5.6% and 5.5% for run - ii and run - iii , respectively . the same values of the systematic uncertainty were adopted to the @xmath19 decay . based on the acceptance for the @xmath19 decay obtained from the mc simulation , and the number of @xmath2 decays in the data - taking runs , the single event sensitivity ( s.e.s ) was defined as @xmath64 the s.e.s . was derived as @xmath65 for run - ii and @xmath66 for run - iii . the combined s.e.s . was @xmath67 . the sensitivity reported here is considerably improved from that of our previous analysis @xcite . in addition to the six - times larger statistics , the acceptance factor was improved for several selection criteria . examples include the adoption of a looser cut for @xmath32 in the two-@xmath21 vertex matching , and optimizing the criteria used to select good photon clusters in the csi calorimeter , both of which were possible in a cleaner environment with less neutron background in run - ii and run - iii . vs. @xmath36 for the bifurcation analysis.,height=204 ] lrrrrrrr region & @xmath68 & @xmath69 & @xmath70 & est . @xmath71 & @xmath72 + signal & 3 & 381 & 2508 & [email protected] & + low @xmath40 & 136 & 6330 & 41418 & [email protected] & 21 & + 2nd low @xmath40 & 1151 & 17525 & 105455 & [email protected] & 229 + high @xmath40 & 0 & 110 & 604 & 0.0 & 1 + 2nd high @xmath40 & 1 & 5 & 41 & [email protected] & 0 + high mass & 6 & 283 & 1086 & [email protected] & 12 + a blind analysis " technique was used to minimize the contribution from backgrounds without examining the candidate events in the signal region . among the background sources , @xmath38 was fully reconstructed and rejected primarily by kinematic variables ( @xmath40 , @xmath36 ) of the @xmath37 system . however , @xmath74 would become a background if two photons were missed due to finite photon - veto inefficiencies in the detector . the kinematic cuts could not be applicable to this dominant background source , and suppression depended strongly on the photon veto conditions . to avoid the difficulty of having enough statistics in the mc simulation of the @xmath43 background in the signal region , we adopted the bifurcation method " to estimate the number of background events in the signal region from the real data @xcite . we imposed several selection criteria ( cuts ) to identify the @xmath75 signal from backgrounds , and the roles of these cuts were categorized in two groups , namely cut @xmath76 and cut @xmath77 . the cut @xmath76 consisted of veto " requirements to ensure no particles other than from the csi - calorimeter region , and the cut @xmath77 consisted of the cuts to select well - reconstructed @xmath21s from pairs of photon clusters in the calorimeter . if the two categories of cuts are uncorrelated , the number of events that passed both of the cuts @xmath76 and @xmath77 would be described as @xmath78 where @xmath79 is the number of events after the basic setup cuts " are imposed prior to the cuts @xmath76 and @xmath77 , and @xmath80 , @xmath81 and @xmath82 are the probability of passing the conditions @xmath76 , @xmath77 , and @xmath83 , respectively . the number of estimated background events @xmath71 is derived as @xmath84 where the symbols @xmath85 and @xmath86 are the inverse logic of @xmath76 and @xmath77 , respectively . thus @xmath71 in the signal region after imposing all of the cuts @xmath76 and @xmath77 could be estimated from the number of events obtained by the combination of @xmath87 , @xmath88 , and @xmath89 . the kinematic region defined by @xmath40 and @xmath36 of the @xmath37 system was further divided into six as shown in fig . [ fig : bif - region ] . for each region , @xmath71 was estimated by the formula above . system for the events with @xmath90 0.1 gev/@xmath42 and @xmath91 0.55 gev/@xmath9 with a loose requirement for @xmath92 for the run - ii and run - iii , assuming the production point was on the downstream edge of cc02 . the solid line is an expectation from the @xmath93 decay.,height=158 ] table [ tab : estbgrun ] summarizes the background estimates for the run - ii and run - iii data combined . various kinematic regions were tabulated to cross - check the validity of this method . the numbers of events of the real data @xmath72 are consistent with the estimated values @xmath71 , and also agree reasonably with the mc results of dominant backgrounds ( sum of @xmath38 and @xmath94 decays ) , after imposing the cuts @xmath76 and @xmath77 ( table [ tab : mcbifurcation ] ) in most of the regions around the signal region . the real data ( 12 events ) in the high mass " region exceeded the estimated number of events ( @xmath95 events ) . these events are seen in fig . [ fig : mcdata](d ) with @xmath36 around @xmath96 , extending to the higher @xmath40 region ( @xmath970.1 gev/@xmath42 ) . [ fig : propertime ] shows the proper - time distribution of the @xmath37 system with @xmath98 0.1 gev/@xmath42 , @xmath99 0.55 gev/@xmath9 , and a looser @xmath33 requirement for the run - ii and run - iii combined , assuming that the production point was the downstream edge of the cc02 counter . the proper - time distribution ( fig . [ fig : propertime ] ) is consistent with that of expected from the @xmath100 decay , implying that the events are from regeneration of @xmath57 in the upstream materials and subsequent @xmath37 decay . because the final state of the @xmath101 process was fully reconstructed , these events were insensitive to the photon - veto cuts ( cuts @xmath76 ) and also to the csi cluster shape cuts ( cuts @xmath77 ) . hence the bifurcation method was not applicable for estimating the backgrounds from this process . nevertheless , these events might occur in the signal region by a mis - pairing of the photons from the @xmath21s . assuming that the mis - pairing rate for @xmath100 decays was equal to that for @xmath38 decays shown in fig . [ fig : mcdata](b ) , the contribution of events in the high mass " region to the signal region was negligibly small ( @xmath102 0.03 events ) . the number of background events in the signal region was estimated to be [email protected] for the run - ii and run - iii combined ( table [ tab : estbgrun ] ) . lrrrrrrr & @xmath103 & @xmath104 & sum of & real + region & ( mc ) & ( mc ) & the two mc & data + signal & 0.0 & 0.0 & 0.0 & + low @xmath40 & 0.5 & 15.6 & 16.1 & 21 + 2nd low @xmath40 & 11.9 & 203.1 & 215.0 & 229 + high @xmath40 & 0.1 & 0.0 & 0.1 & 1 + 2nd high @xmath40 & 0.0 & 0.0 & 0.0 & 0 + high mass & 0.2 & 0.0 & 0.2 & 12 + in the discussions above , we assumed that the cut sets @xmath76 and @xmath77 were uncorrelated . a measure of the correlation was given by the parameter @xmath105 , as the difference of the probability of satisfying the cut @xmath76 for the events that passed the cut @xmath77 , and for the events that passed the cut @xmath86 . corrections to the @xmath71 due to the correlation are given by @xmath106 to first order of @xmath107 @xcite . in the low @xmath40 " region , which was closest to the signal region , @xmath107 was @xmath108 . applying this value to the signal region , the correction to be made to the @xmath71 was @xmath109 events . the magnitude of the correction is small compared with the quoted error for the @xmath71 , and this effect was not taken into account in deriving the final result . vs. @xmath36 distribution for run - ii and run - iii combined . , height=215 ] after all of these studies , the candidate events in the signal region were examined . figure [ fig : opendata ] shows the combined results of the run - ii and run - iii ; no events are observed in the signal region . by using poisson statistics with zero background events in the signal region , we set an upper limit on the branching fraction of @xmath0 to be 8.1@xmath6 at the 90% c.l . the systematic uncertainty in the s.e.s . is not taken into account in the limit . from the same data sample , we also derived the upper limit on the @xmath15 ( @xmath110 invisible particles ) decay assuming three - body phase space . because the acceptance depended on the mass of @xmath8 , the upper limit was obtained as a function of the assumed mass of @xmath8 as summarized in table [ tab : sensitivity ] . lccc @xmath111 & accep- & & br upper limit + ( mev / c@xmath112 ) & tance & s.e.s . & with 90% c.l . + 50 & 3.52@xmath113 & 3.04@xmath6 & 7.0@xmath6 + 75 & 3.50@xmath113 & 3.05@xmath6 & 7.0@xmath6 + 100 & 3.45@xmath113 & 3.10@xmath6 & 7.1@xmath6 + 125 & 2.91@xmath113 & 3.67@xmath6 & 8.5@xmath6 + 150 & 2.07@xmath113 & 5.16@xmath6 & 1.2@xmath114 + 175 & 9.18@xmath7 & 1.16@xmath114 & 2.7@xmath114 + 200 & 6.21@xmath114 & 1.72@xmath7 & 4.0@xmath7 + there are prospects of further improving the limit . a new experiment e14 is now in preparation at j - parc @xcite , aiming at studying the @xmath115 decay with three orders - of - magnitude higher sensitivity than e391a by utilizing higher beam flux , longer beam time , and an upgraded detector with csi calorimeter of higher efficiency and granularity . further increases of the beam flux are planned in the upgrade . in the case of @xmath116 , improving the photon cluster detection in the csi calorimeter is essential to improving the sensitivity . hence we are hopeful to have significant progress on this process in the near future . we are grateful to the operating crew of the kek 12-gev proton synchrotron for the successful beam operation during the experiment . this work has been partly supported by a grant - in - aid from the mext and jsps in japan , a grant from nsc and ministry of education in taiwan , from the u.s . department of energy and from the korea research foundation .	the rare decay @xmath0 was studied with the e391a detector at the kek 12-gev proton synchrotron . based on @xmath1 @xmath2 decays , an upper limit of @xmath3 was obtained for the branching fraction at 90@xmath4 confidence level . we also set a limit on the @xmath5 invisible particles ) process ; the limit on the branching fraction varied from 7.0@xmath6 to 4.0@xmath7 for the mass of @xmath8 ranging from 50 mev/@xmath9 to 200 mev/@xmath9 .
the paradigm of inflation provides nowadays a simple and accurate description of many of the aspects of the universe we observe . it is able to naturally explain several questions like the particle horizon or the flatness problems of early cosmology , among others @xcite . remarkably , this paradigm also explains the origin of the large scale structure . it requires , however , the quantum principles for this mechanism to work @xcite . then , it is one of the most favorable situations where quantum gravity phenomena can potentially be detected . regrettably , the traditional paradigms based on quantum field theories on classical spacetimes are not valid close to the big bang singularity . they simply assume suitable initial conditions at the onset of inflation or during the slow roll phase , where the only relevant quantum phenomena are those coming from small perturbations of matter and geometry . in this manuscript we are interested in the extension of cosmological perturbation theory to those regimes where classical general relativity is not valid anymore . among the different candidates for a quantum theory of gravity , loop quantum gravity ( lqg ) is one of the most developed approaches @xcite . loop quantum cosmology ( lqc ) , the application of lqg techniques to cosmological scenarios , has demonstrated to be a trustworthy formalism @xcite , where the classical singularity is replaced by a quantum bounce , as well as it preserves upon evolution the semiclassicality of quantum states @xcite . this formalism provides an extension of the inflationary scenarios to the planck era , opening the possibility of potentially studying high energy physics and quantum gravity phenomena . in addition , if inhomogeneities are included , even if one follows the formalism of cosmological perturbation theory , the traditional assumptions adopted there must be revisited if one wants to extend this paradigm of the early universe to the deep planck regime . for instance , the classical spacetime approximation might not be valid anymore for non - semiclassical states or close to the high curvature regime of the background geometry . in addition , the criteria that pick out an initial state for the inhomogeneities must also be revised in these new quantum gravity scenarios , since these models admit an extension into the high curvature regime , where general relativity breaks down , but the geometry is still regular ( though not necessarily classical ) . among the different proposals to study cosmological models that include small inhomogeneities in the framework of lqc so far , there is a well known proposal that adopts the so - called _ dressed metric _ approach @xcite . there , the equations of motion of the perturbations evolving on a quantum geometry are influenced by ( not all but some of ) the fluctuations of the background quantum state . this proposal is inspired by the _ hybrid quantization _ @xcite , since it also assumes that the main contributions of quantum geometry are incorporated in the background degrees of freedom , while the inhomogeneities are described by means of a standard representation ( for instance a fock quantization ) . this is a regime between full quantum gravity and quantum field theories on curved spacetimes . the hybrid quantization was originally applied to gowdy cosmologies , providing a consistent and successful quantization free of singularities . afterwards , this quantization was extended to other models , like gowdy cosmologies coupled to matter @xcite . it was also adopted for the study of cosmological inflationary models with small scalar inhomogeneities , where it was provided a full quantization @xcite . however , the dynamics has not been fully understood yet . in ref . @xcite the quantum dynamics was partially solved , assuming a born oppenheimer ansatz for the solutions to the homogeneous scalar constraint , allowing one to recover a dressed metric regime ( under certain approximations ) from a more general formalism . finally , in ref . @xcite , a generally covariant formalism for scalar perturbations has been developed in order to strengthen the predictions of this hybrid approach . there , after a canonical transformation in the phase space ( of the perturbed theory ) , the authors are able to explicitly identify the true physical degrees of freedom ( mukhanov sasaki variables ) without carrying out any gauge fixing . they adopt a born oppenheimer ansatz , derive the dressed metric regime within this fully covariant formalism and provide the effective equations of motion . the genuine ( loop ) quantum dynamics of the background spacetime is a key issue for the dressed metric approach , since all its richness lays in the quantum fluctuations . it is very well known for a free massless scalar field @xcite , but , whenever a potential is added , the dynamics becomes sufficiently intricate that only an effective description has been considered so far @xcite . nevertheless , one can restrict the study to those states for which the dressed metric and the effective dynamics are in agreement ( assuming that they exist ) . within this approximation , the only relevant magnitude to be provided is the value of the scalar field at the bounce ( once its mass has been fixed ) for the homogeneous model . for the inhomogeneities , nevertheless , one specifies the initial data by choosing a suitable vacuum state ( usually ) at that time . however , this is the most important conceptual question to be understood . as we mentioned before , this is a qualitatively different situation with respect to the traditional treatments in which suitable initial data is provided at the onset of inflation or during the slow - roll regime . here , genuine quantum geometry contributions are negligible and then ignored . presently , there is no well understanding of what is their relevance in these new scenarios . in previous studies @xcite in loop quantum cosmology , there have been proposed some candidates , the so - called adiabatic states @xcite , mainly focusing in the fourth order ones . on the one hand , the physical predictions obtained from these adiabatic initial conditions for the perturbations at the bounce seem to be in good agreement with observations whenever the scales for which one obtains large corrections are beyond the observable ones in the cosmic microwave background ( cmb ) . in those situations the main quantum corrections only affect the amplitude of the large scale temperature anisotropies of the cmb , i.e. the range of modes that are either not observable or just entering today the horizon . on the other hand , considering adiabatic states of high order ( fourth order or more ) , a ultraviolet renormalization scheme is available for the stress - energy tensor in order to compute the physical energy density of the perturbations @xcite . nonetheless , it is not completely clear that such family of states ( or equivalently initial conditions ) is the physically preferred choice , at least for those values of the scalar field at the bounce that do not give primordial power spectra where the scales with strong corrections are at the limit of the range of observability or beyond ( usually those that do not produce large number of e - foldings ) . the purpose of this manuscript is to study different primordial power spectra for scalar perturbations that can be obtained from the hybrid quantized model commented above @xcite , and qualitatively confront the obtained physical predictions with other approaches in loop quantum cosmology , with particular attention to the dressed metric approach , and with observations provided by planck mission @xcite . we will assume that the states of the background are highly peaked on the effective geometry , so that it is in good agreement with the dressed metric regime of the hybrid approach . this consideration allows us to determine the dynamics of the background and the perturbations by means of a set of differential equations that incorporate quantum geometry corrections . for convenience , we will consider initial data at the bounce , since the background dynamics is fully determined by one parameter : the value of the homogeneous mode of the scalar field . we will consider as initial state of the perturbation the adiabatic states studied in refs . @xcite and @xcite of 0th , 2nd and 4th order . for them , we found that the primordial power spectrum at the end of inflation agrees with observations if there is sufficient e - foldings , but none of the adiabatic states considered in this manuscript provides a suppression of the power spectrum , but an enhancement , for the observable large scale modes . in addition , in order to face the question of the vacuum state of the perturbations at the bounce , i.e. their initial conditions , we have investigated an alternative prescription based on a genuine algorithm that computes numerically an initial vacuum state in such a way that the time variation of the amplitude of the primordial power spectrum of the mukhanov sasaki variable from the bounce to the end of the kinematically dominated epoch is relieved . this method provides a vacuum state at the bounce that produces a primordial power spectrum at the end of inflation that is in good agreement with the current observations of planck mission . remarkably , it is suppressed at large scales , as it seems to be favored by current observations . furthermore , there is an important contrast with respect to the predictions of the adiabatic states since it does not present the highly oscillating region and averaged enhancement at large scales typical of those states in this scenario . this paper is organized as follows . in section [ sec : eff - dyn ] we specify the ( semi)classical setting and its dynamics by means of a set of effective equations of motion within the hybrid quantization approach . the initial value problem is studied in section [ sec : init - state ] , where we consider several choices of initial vacuum state for scalar perturbations . there we provide a new constructive method to select a suitable initial vacuum state . we also consider adiabatic states to different order . in section [ sec : ps - atps ] we compute the primordial power spectrum at the end of inflation for these vacuum states . we also provide several examples of the power spectrum of temperature anisotropies for the new vacuum state and compare with observations . we conclude and discuss our results in section [ sec : dis - conc ] . for the sake of completeness , we include an appendix where the previous constructive method is applied to a particular quantum field theory in a de sitter space . the system we will study here is a flat friedmann robertson walker spacetime with @xmath0 topology . it is endowed with a spacetime metric characterized by a homogeneous lapse @xmath1 and a scale factor @xmath2 multiplying the auxiliary three - metric @xmath3 of the three - torus . the angular coordinates in this spatial manifold are @xmath4 such that @xmath5 , where @xmath6 is the period in each orthonormal direction ( for simplicity the three periods have been chosen to be equal ) . the auxiliary three - metric @xmath3 induces a laplace - beltrami operator whose eigenfunctions @xmath7 and eigenvalues @xmath8 are well known ( see e.g. ref . @xcite ) . here @xmath9 is any tuple whose first component is a strictly positive integer . we will adopt a description of this homogeneous geometry in terms of lqg variables . there , one starts with an su(2)-connection and a densitized triad as basic canonical variables . however , the connection itself is not well defined in the full quantum theory but holonomies of the connection . besides , in the context of lqc , we will adhere to the improved dynamics scheme @xcite . this choice is motivated by the fact that such cosmologies bounce whenever the energy density achieves a universal critical value @xmath10 for semiclassical states , where @xmath11 is the planck energy density . in this situation , the classical homogeneous canonical variables of the geometry are the volume @xmath12 and the canonically conjugated variable @xmath13 ( which is proportional to the hubble parameter ) . these variables satisfy the classical algebra @xmath14 , where @xmath15 is the newton constant and @xmath16 the immirzi parameter @xcite . we will couple to this model a massive scalar field @xmath17 fulfilling , with its canonical momentum , @xmath18 . following the analysis of ref . @xcite , we will introduce scalar perturbations around the classical homogeneous variables . this was done in a closely related scenario in a seminal paper by halliwell and hawking @xcite . the perturbative expansion of the einstein - hilbert action of general relativity coupled to a massive scalar field is truncated to second order . the resulting action , in the canonical formalism , can be written in terms of a hamiltonian that is a linear combination of constraints : the homogeneous mode of the scalar constraint that contains quadratic contributions of the perturbations , one hamiltonian and one diffeomorphism constraints , both ( local and ) linear in the perturbations . these two last constraints generate the so - called gauge transformations of the perturbations . the common strategy followed in the theory of cosmological perturbations is to work with gauge invariant potentials @xcite . in this sense , in ref . @xcite it was proposed to implement the gauge invariant potentials at the action itself by means of the hamilton - jacobi theory . however , the explicit dependence of the hamiltonian in terms of the non gauge - invariant variables was ignored as well as the background variables were not properly transformed . these last steps were completed in ref . @xcite . following it , we consider a canonical transformation involving the whole phase space , in order to separate the gauge invariant mukhanov sasaki potential @xmath19 , and its conjugated momentum @xmath20 , with respect to the remaining gauge variables , that will be defined as @xmath21 , for @xmath22 . the details can be found in ref . the total hamiltonian , to second order in the perturbations , will be of the form @xmath23 where @xmath24 and @xmath25 are the fourier modes of the inhomogeneous lagrange multipliers associated with the constraints linear in the perturbations of the final action . we will then adopt a hybrid quantization approach , combining a loop quantization for the homogeneous connection , and a standard representation for the scalar field and the inhomogeneities @xcite . since the model involves first class constraints , we will adopt the dirac quantization approach . the physical states must be annihilated by the operators corresponding to the constraints . in this case , one can easily realize that any physical state must be independent of @xmath26 , for @xmath22 . the remaining condition to be imposed is that physical states must be annihilated by the homogeneous mode of the hamiltonian constraint , i.e. by @xmath27 . since we do not know yet how to solve this quantum constraint , we will compute approximate solutions by adopting first a born oppenheimer ansatz for them , and then additional approximations ( see refs . @xcite for further details ) that are expected to be fulfilled for instance by semiclassical states with small dispersions . in addition , among them we choose those ones where the effective equations of motion of the expectation values of the basic observables agree with the effective dynamics of loop quantum cosmology ( though we do not know yet if this family of states really exists , it seems natural to assume it does ) . this effective dynamics has been studied in a similar scenario @xcite corresponding to a flat frw spacetime filled with a massless scalar field . the effective equations of motion provided there approximate very well the exact quantum evolution of the expectation values of quantum operators for semiclassical states . following this motivation , the effective hamiltonian constraint that we will employ in this manuscript is obtained from the quantum hamiltonian constraint @xmath28 by replacing expectation values of products of operators by products of expectation values of such operators . under these assumptions , the dynamics is generated by the effective hamiltonian constraint @xmath29 with the unperturbed hamiltonian constraint @xmath30 in the last expression , @xmath31 denotes the mass of the massive scalar field and @xmath32 the minimum non - zero eigenvalue of the area operator in lqg . besides , @xmath33 is the quadratic contribution of each of the modes @xmath34 with @xmath35 where @xmath36 and @xmath37 the above expressions are easily obtained from refs . @xcite by using the previously mentioned effective dynamics prescription . it is worth commenting that , for convenience , one can replace @xmath38 by @xmath39 in . this is in agreement with our approximate effective equations of motion and the perturbative scheme we adopt . we have also neglected corrections of the inverse of the volume operator @xmath40~\simeq~1/v+{\cal o}(v^{-2})$].the effective equations of motion for the phase space variables are obtained by computing their poisson brackets with the effective hamiltonian constraint @xmath41 given in the previous expressions . for the background variables the equations of motion are given by [ eq : hom - eqs2 ] @xmath42 where @xmath43 , for any phase space variable @xmath44 . for the sake of completeness we have included backreaction contributions , but for all practical purposes we will neglect them in the following . as a consequence , the effective equations of motion for the background will be equivalent to the ones used in ref . ( 2.9 ) , ( 2.11 ) and ( 4.2). the time evolution of the perturbations , on the other hand , is dictated by an infinite number of first order differential equations : @xmath45 which do not mix different modes and ( given our perturbative truncation ) are linear in the inhomogeneities . in the following , we will only consider conformal time @xmath46 , which involves @xmath47 . in addition , and for the sake of simplicity , we will set @xmath48 but with @xmath49 taking continuous values . this is in agreement with the approximation @xmath50 . it is enough for all practical purposes to consider @xmath6 much bigger than the hubble radius during the whole evolution . in standard inflationary cosmology it is common to give initial data at the onset of inflation for both the background and the perturbations . one can also consider initial data close to the classical singularity , but keeping in mind that einstein s theory breaks down there and any physical prediction can not be fully trusted . nevertheless , in that regime , one expects that quantum gravity effects will become relevant , and the emergent preinflationary scenario can potentially change the traditional picture . this is indeed the case of our model . in the present scenario the classical singularity is replaced by a quantum bounce whenever the energy density of the system reaches a critical value @xmath51 of the order of the planck energy density . in lqc usually one considers initial data when the energy density reaches @xmath51 . this is so because there the hubble parameter vanishes , the value of the scale factor is chosen equal to the unit ( its concrete value is in fact irrelevant for open flat topology ) , and the momentum conjugated to the field is determined by the scalar constraint ( i.e. the friedmann equation ) . therefore , any solution to the effective equations of motion is completely determined by the value of the homogeneous mode of the scalar field at the bounce , i.e. @xmath52 . in conclusion , concerning initial conditions of the background variables , we consider @xmath53 as the only free dynamical parameter . in addition , one can also allow for different values for the mass of the scalar field @xmath31 , and then consider it as an additional ( homogeneous ) free parameter . however , the specification of initial data for the perturbations is not fully understood yet , though there are several natural candidates as initial state of the inhomogeneous sector . this freedom in the choice of initial vacua is tantamount to the fact that quantum field theories in general curved spacetimes do not possess a sufficient number of symmetries allowing to choose a unique vacuum state @xcite . the best one can do , by now , is to assume additional physical or mathematical criteria that select a given candidate among all possible choices . but let us first remind that the selection of a fock vacuum state is equivalent to select a complete set of creation and annihilation variables . in turn , this is equivalent to define a complete set of `` positive frequency '' complex solutions @xmath54 to the equation of motion that satisfy the normalization relation @xmath55 in the last equation the prime stands for derivation with respect to conformal time whereas the asterisk stands for complex conjugation . since the equations of motion for the mukhanov sasaki variables are linear , the different choices of sets of `` positive frequency '' solutions can be translated to different choices of initial conditions @xmath56 at the considered initial time @xmath57 . also the normalization relation is preserved upon evolution and therefore it is only necessary that such condition is satisfied initially . general initial conditions ( up to a irrelevant global phase ) can be written as @xmath58 where @xmath59 is a non - negative function of the mode , @xmath49 , whereas @xmath60 is an arbitrary real function . one can restrict the ultraviolet behavior of those functions using well motivated physical and mathematical conditions . for instance , for compact spatial slices one can impose unitary evolution ( in addition to invariance under spatial symmetries ) to the fock quantization @xcite . extending this criterion to open spatial sections , for instance by choosing a vacuum with a finite particle production per spatial volume upon evolution , one can see that the previous functions must behave like @xmath61 and @xmath62 for large @xmath49 , with @xmath63 . other well known and widely used criteria to select suitable vacuum states , as the hadamard condition @xcite or the adiabatic states @xcite , give a similar ultraviolet restriction @xcite . we will restrict our study to those initial conditions satisfying the unitary evolution requirement . nonetheless , since this requirement only constraints the ultraviolet behavior , there still exists an infinite freedom . we then need to further restrict the possible candidates . one of the most popular ways to constraint or select a set of `` positive frequency '' solutions ( or initial conditions ) is by means of the so - called adiabatic states . in order to define a complete set of complex solutions for them one considers the following form @xmath64 if this expression is introduced in the equations of motion one obtains for the function @xmath65 the following equation @xmath66 here , @xmath67 stands for the corresponding time - dependent mass term which in our case is given by @xmath68 adiabatic solutions are then obtained by choosing conveniently functions @xmath69 that provide approximations to the exact solution to equation that converge to them at least as @xmath70 in the limit of infinite large @xmath49 , with @xmath71 being the _ order _ of the adiabatic solution where the different approximate solutions , i.e. the adiabatic orders , are determined in the limit @xmath72 . nonetheless , it is well known that this is essentially equivalent to consider the limit of infinitely large @xmath49 . ] . it is important to remark that those adiabatic solutions are suitable approximated solutions in the limit of large @xmath49 but it is also well known that they do not provide in general a good approximation for small @xmath49 . they were originally introduced with the purpose of adopting analytical tools in several situations , and later on it was shown they are also useful for other purposes like renormalization of the stress - energy tensor in cosmological scenarios . however , we are not interested in analytical approximations to the exact solutions but instead in using these adiabatic states to obtain suitable initial conditions for our equations of motion , as it was done in ref . @xcite for the dressed metric approach , in order to study the physical predictions of the hybrid quantization approach , in particular , as well as the robustness of loop quantum cosmology , in general . actually , from equation one can define initial conditions associated to an adiabatic solution ( absorbing an irrelevant global phase ) by choosing @xmath73 and @xmath74 , with both expressions in the right hand sides evaluated at @xmath75 . now , we are going to introduce two different procedures that we will use in this work to obtain specific adiabatic initial conditions of different orders . the first procedure mimics the one explained in the reference @xcite . one obtains an adiabatic solution of order @xmath76 , @xmath77 , by plugging in the right hand side of the equation the @xmath78 order solution @xmath69 , starting with @xmath79 ( the solution naturally associated to a free massless scalar field ) . the other procedure that we use to select specific adiabatic initial conditions for each order is to consider the above mentioned solutions and perform an asymptotic expansion in the limit of large @xmath49 and then truncating it at the considered order . this method is equivalent to the one used in ref . the functions obtained by means of this last procedure will be denoted as @xmath80 . in the following we will consider these two specific adiabatic initial conditions until 4th order . to show some instances of the functions considered here , obviously both cases lead to the same initial conditions for the 0th order , @xmath81 , whereas for the 2nd order we have @xmath82 and @xmath83 . it is worth noting that it is not guaranteed that these two procedures give meaningful initial conditions ( and therefore solutions ) since they can lead to functions @xmath69 that are negative or even complex for some values of @xmath49 . in addition to the above mentioned potential problems of the two procedures to obtain initial conditions associated to adiabatic states it is easy to realize that , in general , both of them give different sets of `` positive frequency '' solutions when considering different initial times . this may not be seen as a big drawback but it is important to stress that , for instance , in de sitter spacetimes the privileged bunch davies vacuum is selected by one specific set of solutions irrespectively to the time selected to define the corresponding initial conditions . in summary , these adiabatic conditions allow us to reduce the possible initial states for the inhomogeneities , but from the definition of adiabatic states , it is clear that there still exist infinitely many choices of adiabatic states of any order . recently , it has been proposed a new criterion to select a unique set of initial conditions in the context of adiabatic vacua @xcite . the criterion selects those initial conditions for which the expectation value of the adiabatically regularized stress - energy tensor at the initial time vanishes . besides the previous prescriptions for the choice of initial data , in this work we will also propose a criterion to select a set of initial conditions , but based on different arguments . it focuses on the behavior of the relevant quantity in our computations : the primordial power spectrum of the comoving curvature perturbation @xmath84 . the primordial power spectrum @xmath85 is defined from its 2-point function ( which is in one - to - one relation with a concrete vacuum state ) as @xcite @xmath86 where @xmath87 is the conformal time at which the quantities in the expression above are evaluated ( usually at the end of inflation ) . it can be written , in terms of the `` positive frequency '' solutions of the gauge invariant mukhanov sasaki potential , as @xmath88 where @xmath89 is given in terms of the hubble parameter defined as @xmath90 . our criterion is based on selecting the initial conditions that minimize the temporal variation of @xmath91 in a selected time interval . they can be obtained as follows . for each mode , we define the quantity @xmath92 that clearly depends on the specific solution @xmath93 , and therefore on the initial conditions , and also in the considered temporal integration limits . then , the initial conditions we are looking for are the ones that minimize , i.e. , we pick out the specific values for the real variables @xmath59 and @xmath94 that minimize @xmath95 . the idea behind this constructive criterion is to select solutions for which the quantity @xmath96 does not oscillate rapidly _ and _ with large amplitudes . this choice is justified by the fact that other natural vacuum states in quantum field theories , like the poincar - invariant vacuum for time - independent scenarios or the bunch davies vacuum for de sitter spacetimes , are absent of such fast ( and large ) time - dependent oscillations we show a particular example where we recover the bunch - davies state out of this algorithm . ] . they are reflected in the corresponding primordial power spectra , showing strong oscillations in @xmath49 . we understand that this behavior should be weakened if one is selecting a more suitable vacuum state ( at least approximately ) . the primordial power spectrum obtained by means of this method is non - highly oscillating and provides the lowest power when averaging in smalls bins of @xmath49 for @xmath97 ( among the states considered in our study ) . it is worth comenting that , as we show in app . [ app : bd - vac ] , this criterion allows us to identify the bunch - davies vacuum state of a massless scalar field on a de sitter spacetime . therefore , we will regard it as a good choice of initial data . in this work we will take the initial time @xmath98 as the time of the bounce @xmath99 , which is the one in which we are giving the initial conditions . furthermore , we set @xmath100 as the time in which @xmath101 vanishes for the first time from the bounce , which can be considered as the beginning of the inflationary phase ( do not confuse with the slow roll regime ) . in the following ( clearly abusing of the language ) we will refer to the initial conditions obtained in this way as the `` non - oscillating '' initial conditions . nonetheless , it is important to note that , since the time interval considered is finite , this criterion is not expected to give such desirable non - oscillating solutions ( in @xmath96 ) for @xmath102 . in figure [ fig : dk - ck ] we compare the functions @xmath59 and @xmath60 for the different adiabatic initial conditions considered here as well as the `` non - oscillating '' ones . as we can see , in the case of @xmath59 , all choices give the same behavior for @xmath103 , but they strongly differ otherwise . for the functions @xmath60 we see that the `` non - oscillating '' initial conditions and the adiabatic vacuum ones differ in the shown range of @xmath49 , although all of them go quickly to zero for large values of @xmath49 . and left graph : comparison of the @xmath59 function . right graph : comparison of the @xmath60 functions . here the dashed lines correspond to negative values of @xmath60 . note that for the 0th order adiabatic initial conditions @xmath105 . , title="fig:",scaledwidth=49.0% ] and @xmath104 . left graph : comparison of the @xmath59 function . right graph : comparison of the @xmath60 functions . here the dashed lines correspond to negative values of @xmath60 . note that for the 0th order adiabatic initial conditions @xmath105 . , title="fig:",scaledwidth=49.0% ] the `` non - oscillating '' initial conditions , as well as the considered adiabatic initial data , are almost independent of the values chosen for @xmath53 and @xmath31 in the ranges that are physically interesting . this is easy to understand taking into account that the actual values of @xmath53 and @xmath31 are almost irrelevant in the dynamics of the superinflationary phase after the bounce and the kinematically dominated phase . in order to test potential predictions of our proposal regarding the planck era that can be observed in the cosmic microwave background , we study the primordial power spectrum of the comoving curvature perturbations at the end of inflation see equation . this magnitude is indirectly related with observations through the angular spectrum of temperature anisotropies of the cmb . for any pair of homogeneous initial conditions , i.e. , the mass of the field and value of the field at the bounce , we first compute the primordial power spectrum for the comoving curvature for the `` non - oscillating '' initial conditions , @xmath106 , which is going to be considered as the reference one . such primordial power spectrum is obtained from the set of complex solutions @xmath107 , that in the following we are going to denote simply as @xmath108 . any other set of complex solutions is related with the non - oscillatory one by means of @xmath109 with @xmath110 . these mode - dependent complex coefficients @xmath111 and @xmath112 are completely determined by the particular initial conditions @xmath113 under consideration and the `` non - oscillatory '' ones @xmath114 . actually , @xmath115 , \quad \beta_{k}=-i\left[-u^{\prime}_{k,0}v_{k,0}+u_{k,0}v^{\prime}_{k,0}\right].\ ] ] then , one can write the squared absolute value of the new set of complex solutions , and hence the primordial power spectrum , in terms of the reference one as @xmath116)\|u_{k}\|^{2},\ ] ] where @xmath117 $ ] denotes the real part . since we are considering as the reference set of solutions the `` non - oscillatory '' ones , and taking into account the behavior of the initial conditions considered in the previous section , it is easy to realize that the term producing time oscillations for @xmath96 is the one containing the real part of @xmath118 . therefore , for every set of initial conditions we will consider as well the primordial power spectrum obtained by removing that oscillatory part @xmath119 it is worth to comment that for @xmath120 , where @xmath121 denotes the conformal time from the bounce until the end of inflation . for the instance with @xmath122 and @xmath123 , the beginning of inflation ( defined as the moment in which @xmath101 equals to zero for the first time ) occurs at @xmath124 , the slow - roll regime starts at @xmath125 and the inflationary era ends at @xmath126 . for other values of @xmath53 and @xmath31 , in the ranges explored , these times take similar values . also we have selected the units to perform numerical simulations such that @xmath127 = 1 . ] , @xmath128 will give a good approximation of the averaged power spectrum @xmath129 with respect to small bins , whose size in @xmath49 ( or even in @xmath130 ) must be selected in order average the oscillations ( which are expected anyway to be unobservable in the present observations of the cmb ) . in figure [ fig : ppss ] , we compare the primordial power spectra obtained for the initial conditions considered in the previous section , i.e. , the `` non - oscillatory '' ones , the free massless initial conditions and the two sets of 2nd and 4th adiabatic initial conditions . also we compare those primordial power spectra with the one obtained with the first order slow roll formula @xmath131 where @xmath132 is the hubble parameter , @xmath133 is the first slow roll parameter in the hubble flow functions , the subscript @xmath134 means that the quantity is evaluated when the mode crosses the hubble horizon @xmath135 and the dot stands for the derivative with respect to the proper time . and @xmath104.,title="fig:",scaledwidth=49.0% ] and @xmath104.,title="fig:",scaledwidth=49.0% ] + and @xmath104.,title="fig:",scaledwidth=49.0% ] and @xmath104.,title="fig:",scaledwidth=49.0% ] + and @xmath104.,title="fig:",scaledwidth=49.0% ] and @xmath104.,title="fig:",scaledwidth=49.0% ] as we can see in the first graph of figure [ fig : ppss ] the power spectra obtained from the non - oscillatory initial conditions and the slow - roll formula are almost equivalent for modes that cross the hubble horizon when the slow - roll approximation is valid and @xmath136 is less than @xmath137 . ] . nonetheless , we obtain slightly more power in such region for the `` non - oscillatory '' initial conditions because the slow - roll formula is indeed an approximation that is not able ( at the considered order ) to achieve the required precision . on the other hand , for modes that exit the hubble horizon before the slow - roll approximation starts to be valid we obtain that the non - oscillating spectrum predicts less power than the slow - roll formula and shows two regions with different behaviors : ( i ) small oscillations for @xmath138 and ( ii ) strong power suppression for @xmath139 . in the rest of the graphs in figure [ fig : ppss ] we show the primordial power spectrum obtained for the adiabatic initial conditions . one can see that for all of them one obtains a highly oscillatory primordial power spectrum . we can distinguish three regions with different behavior . for large @xmath49 the amplitude of the oscillations tends to vanish and therefore it gets a behavior similar to the `` non - oscillatory '' spectrum and the slow roll formula . for small @xmath49 , one gets a suppression of power for @xmath140 , @xmath141 , @xmath142 and @xmath143 and a large constant power for @xmath144 in the limit @xmath145 . in the intermediate region all the spectra show high oscillations with an averaged enhanced power . it is important to note that these large enhancements are not compatible with current observations unless such region correspond to scales that are not currently observable in the cmb . with the aim at comparing the obtained primordial power spectra with the ones preferred by the observations and statistical analysis of planck mission one has to perform first a _ scale matching_. in fact , we have set arbitrarily @xmath146 whereas observationally the scale factor is usually set as the unit nowadays ( @xmath147 ) . then , in order to obtain a correspondence of the comoving scales @xmath49 with the ( physical ) observational ones we ( naively ) match them using the observational power amplitude @xmath148 at some pivot mode @xmath149 . we will use here the pivot scale used by the planck mission @xcite @xmath150 and the amplitude obtained for the best fit . ] of the tt+lowp data given by @xmath151 . therefore , we will define our pivot comoving scale @xmath149 as the one such that @xmath152 and such that it exits the hubble horizon in ( or closer to ) the slow - roll region . the value of the pivot scale @xmath149 obtained in this way depends both on the mass of the scalar field @xmath31 and the value of its homogeneous mode at the bounce @xmath53 . once such scale matching is done we compare the primordial power spectrum of the `` non - oscillatory '' solutions that we suggest in this manuscript with three parametrized primordial power spectra studied by the planck collaboration . the first one is the simple power - law @xmath153 which is parametrized by the power amplitude @xmath148 at the pivot scale and the spectral index @xmath154 . the planck collaboration obtains from the tt+lowp data that @xmath155 . this simple form of the primordial power spectrum is the statistically preferred by planck , not because it provides the best fit to the observational data , but because it yields a good fit with a remarkably small number of parameters . actually , a better fit to the planck data is obtained when one allows the primordial power spectrum to deviate from the simple power - law form , either considering a running of the spectral index or different functional forms . such improvement of the fitting is mainly due to the possibility of suppressing the primordial power spectrum at large scales . in order to obtain a better fit there , the planck collaboration considers two forms for the primordial power spectra that include two additional parameters . the first one consists in a simple power - law spectrum multiplied by an exponential cut - off : @xmath156\right\}.\ ] ] this power spectrum is typical in scenarios in which slow roll is preceded by a stage of kinetic energy domination @xcite . the second one is a broken - power - law ( bpl ) potential of the form @xmath157 with @xmath158 to ensure continuity at @xmath159 . the best fit to the tt+lowp planck data for the first model is given by @xmath160 , @xmath161 and @xmath162 . for the second model the best fit is obtained with @xmath163 , @xmath164 and @xmath165 . both of them give a slightly better fitting for the observational data , although not good enough to be statistically preferred over the simple power - law spectrum @xcite with two parameters less . in figure [ fig : ppsskp ] we compare the primordial power spectra for the `` non - oscillatory '' initial vacuum for different values of the homogeneous initial conditions with planck best fit of the previous parameterized power spectra . . left graph : fixed @xmath166 . the vertical dashed line corresponds to the largest scale observable in the cmb.,title="fig:",scaledwidth=49.0% ] . left graph : fixed @xmath166 . the vertical dashed line corresponds to the largest scale observable in the cmb.,title="fig:",scaledwidth=49.0% ] + as we see our proposal has a large scale power suppression that resembles the one given by the broken - power - law best fit of planck for scales observed in the cosmic microwave background . finally , we can see the qualitative consequences of the `` non - oscillatory '' initial conditions at the bounce in the power spectrum of temperature anisotropies . we have carried out the computation employing the code @xcite and the cosmological parameters provided by the planck collaboration @xcite for the best fits of both the simple power - law and cut - off models @xcite . in fig . [ fig : anist ] we plot the planck collaboration observational data for the temperature correlations , the planck best fit and the predictions of the primordial power spectra obtained for the `` non - oscillating '' initial conditions . we choose a particular value of the mass of the scalar field equal to @xmath167 and several choices of its homogeneous mode at the bounce . we observe that our `` non - oscillating '' initial conditions can explain the power suppression for small angular momenta ( large cosmological scales ) of the anisotropies of the cmb . this suppression is stronger in those cases where there is not enough inflation , but for a sufficiently high number of e - foldings we still recover the simple power - law primordial power spectrum statistically preferred by planck . for the left graph in fig . [ fig : anist ] we have used the best - fit cosmological parameters provided by the planck collaboration whereas for the right graph we use the cosmological parameters for the best fit of cut - off models @xcite . we observe that the peaks of the baryonic resonances are quite sensible to the two sets of cosmological parameters provided by planck collaboration , mainly because the values of @xmath52 and @xmath31 must be accurately selected according to the change of the matching scale at the pivot mode , but in both cases they still share the same qualitative properties . indeed we have not carried out a rigorous statistical analysis in order to determine the best fit parameters for the primordial power spectrum predicted by the `` non - oscillating '' initial conditions . it will be a matter of future research . it is also interesting to notice that it seems that our `` non - oscillating '' vacuum produces a strong suppression of the primordial power spectrum that is not able to explain the anomalies around @xmath168 , without strongly suppressing the power spectrum of temperature anisotropies at larger scales . although we have not carried out a detailed analysis , one possible explanation is that the small oscillations that precede the strong suppression of the primordial power spectrum of the `` non - oscillating '' vacuum be responsible of those anomalies if the amplitude of these oscillations is big enough . however , it is not clear to us by now which physical process could be behind it . in this manuscript we have computed the primordial power spectrum of the comoving curvature perturbation in the framework of hybrid loop quantum cosmology . since the genuine quantum dynamics has not been fully solved yet , we have considered the effective equations coming from such hybrid quantization ( simply replacing operators by expectation values ) . additionally , we have neglected the backreaction of the perturbations to the background dynamics . it simplifies the dynamics and allows us to compare the results of the hybrid and the dressed metric approaches . with these assumptions we have computed the primordial power spectrum obtained for different choices of initial vacuum states at the time of the bounce for the mukhanov sasaki variables . more specifically , we have considered two different procedures to obtain specific adiabatic - like initial conditions of arbitrary order and computed the primordial power spectra for 0th , 2nd and 4th orders . they are in good agreement with the ones obtained within the dressed metric approach @xcite . therefore , it is remarkable that the predictions of loop quantum cosmology seem to be robust , since these two formalisms , constructed following different strategies , provide qualitatively similar predictions under the same physical conditions . note that , although we have neglected backreaction contributions and consider the lqc effective dynamics , the effective equations of motion of the perturbation are different in the hybrid and the dressed metric approaches mainly due to the way in which polymeric corrections are included . the consequence is that the quantitative final results will be different , as well as the adiabatic - like initial data in the two approaches will not agree since they involve time - dependent functions that do not coincide when quantum gravity corrections are important ( i.e. , at the bounce ) . the primordial power spectra for adiabatic vacuum states in the hybrid and the dressed metric approaches have in common three distinct behaviors at different scales : ( i ) smooth slow - roll - like behavior ( with small oscillations ) for @xmath169 , ( ii ) large oscillations with an averaged power enhancement for @xmath170 and ( iii ) strong power suppression for @xmath139 , except for the expanded adiabatic vacuum of order four @xmath171 for which the power remains large and constant , at least in the hybrid approach . it is worth to mention that this kind of primordial power spectra , though they can be in good agreement with observations if the most important corrections correspond to scales that are not currently observable in the cmb , are not able to successfully explain the suppression of the temperature anisotropy power spectrum at large scales , without further considerations @xcite . let us also comment that , at first glance , neither our results nor the ones of the dressed metric proposal are in agreement with the ones obtained within the _ deformed algebra _ approach , that leads to highly oscillatory ( with large amplitude ) primordial power spectra even at small scales @xcite . in addition to adiabatic - like initial conditions for the perturbations at the bounce , we have also provided a new criterion to select a suitable initial vacuum state . such criterion picks out the initial conditions for each mode in such a way that it minimizes the time variation of the amplitude of the mukhanov sasaki variable from the bounce to the beginning of inflation . we have shown that such `` non - oscillating '' initial conditions yield a primordial power spectrum where the large oscillations with the averaged enhanced power are not present , obtaining instead for that range of scales a behavior compatible with the one obtained from the slow - roll formula . remarkably , it presents a strong power suppression for large scales . consequently , it may provide a better fitting to the current observations than the spectrum obtained from a quadratic potential considering only the slow - roll regime or from the usual simple power - law primordial power spectrum . nonetheless , although a rigorous statistical analysis is necessary , it seems that the obtained strong power suppression is not enough to explain the observed anomaly around the multipole @xmath172 in the temperature anisotropy angular spectrum . one appealing possibility is to consider initial conditions for the mukhanov sasaki variables that slightly deviate from the ones obtained with the considered criterion . such initial conditions would lead to a primordial power spectrum with slightly larger oscillations around @xmath173 that might explain the above mentioned anomaly . those slightly differently initial conditions might be obtained by different physical processes . one possibility is , for instance , by minimizing the time integrated ( in conformal time ) value of the energy density for each mode from the bounce to the beginning of inflation . on the other hand , since we expect that the set of complex solutions selected by `` non - oscillating '' initial conditions give approximately similar physical results to the ones obtained by giving minkowski - like initial conditions in the kinematically dominated period , another possible way to obtain a primordial power spectrum with larger ( but still small ) oscillations around @xmath173 is defining minkowski - like initial conditions around the end of the superinflationary era after the bounce . indeed , it is not clear to us that the quantum dynamics of the universe is fully determined by our approach but , instead , there is a decoupling scale right after the bounce where the hybrid quantization ( and so the dressed metric approach ) is valid and where it is natural to give minkowski - like initial conditions . finally , another possibility is to break the hypothesis of isotropy before inflation @xcite , being the anisotropies the main source of oscillations at scales of the order of @xmath173 . the assumption of isotropy of the universe together with the truncation of the perturbations to second order in the action allow us to focus our attention on scalar perturbations , since they decouple dynamically from the vector and tensor modes . in these circumstances , the vector inhomogeneities are non - dynamical degrees of freedom . however , the tensor modes can not be neglected if one wants a complete physical picture of the system ( under the previous hypotheses ) . although the hybrid quantization approach is still incomplete at this respect , our preliminary calculations suggest that it is possible to incorporate tensor modes in this formalism without further considerations . the effective equations of motion are well defined at the planck regime and have a well behaved ultraviolet limit . our purpose in the future is to compute the tensor primordial spectra for several initial adiabatic vacuum states ( as well as for the `` non - oscillatory '' one ) within this hybrid quantization approach . although it is soon to draw any conclusion within this formalism , the analyses by planck collaboration suggest that a quadratic potential is statistically disfavored since it predicts a tensor - to - scalar ratio slightly higher than the upper bound @xmath174 ( 95%cl ) . however , whether this is true in the hybrid formalism and the possible physical phenomena that could deal with this question is something that we will discuss in a forthcoming publication . our results are mainly based on the selection of the `` non - oscillating '' vacuum state , which has been obtained by following a particular algorithm whose main purpose is to minimize the oscillations of the primordial power spectrum . but additional considerations can be further investigated . for instance , the quantity inside the integral in eq . is the absolute value of the derivative of @xmath96 with respect to conformal time . although the algorithm works very well eliminating most of the oscillations in this physical quantity ( let us recall that it is related with the 2-point function ) , one could instead consider minimizing these type of oscillations in other physical quantities like either the total , kinetic or potential time - dependent energy of each mode . besides , it would be interesting to apply this algorithm to different cosmological settings . in particular , there are some situations where this method can be tested since the natural vacuum state is already known . this is the case , for instance , of time - independent scenarios or de sitter spacetimes . if we consider arbitrary initial conditions , like in eq . , it would be very interesting to see if this algorithm is able to reconstruct the privileged initial conditions of the poincar - invariant vacuum or bunch - davies vacuum , respectively . a preliminary study for a test , massive scalar field in a minkowski spacetime shows that the algorithm is able to find in a good approximation the initial conditions for the poincar - invariant vacuum state . all these aspects will be addressed in future publications . in summary , the study carried out in this manuscript provides novel ideas about the extension of the traditional inflationary paradigm of cosmological perturbations theory to the planck era . when the big bang singularity is avoided , and the evolution of the universe can be extended far in the past with respect to the onset of inflation , it is natural to ask again these two important questions : i ) what is the natural initial state of the universe in the past , for instance , at the high curvature regime ? and ii ) how predictive are these new scenarios with respect to the present observations ? we show here that loop quantum cosmology and the hybrid quantization approach of this particular model suggest a possible answer to these questions . besides , the usually ignored freedom about the choice of initial state of the perturbations has been considered in this manuscript . we have provided a new criterion that can seed light on the understanding about the existence of privileged vacuum states that has not been considered before in the way we do here , at least to the knowledge of the authors . we also strongly believe that our criterion is not restricted to scenarios in genuine quantum cosmology but they can also be adopted in many other models in cosmology without further considerations . the authors are greatly thankful to i. agull , l. castell gomar , m. martn benito , g. a. mena marugn and t. pawowski for enlightening conversations and suggestions reflected in the manuscript . we also thank j. torrado for his clarifications about the code . this work was supported in part by pedeciba , and the grants micinn / mineco fis2011 - 30145-c03 - 02 and fis2014 - 54800-c2 - 2-p from spain . d. m - db is supported by the project conicyt / fondecyt / postdoctorado/3140409 from chile . j. o. acknowledges support by the grant nsf - phy-1305000 ( usa ) . in this appendix we will show that the criterion of minimizing ( mode by mode ) the time variation of the power spectrum , introduced in the section [ sec : init - state ] , allows us to pick out the bunch - davies state when considering a test scalar field in a de sitter spacetime . we will restrict the study to a massless scalar field in the cosmological chart of this spacetime , as it was done in ref . we will first briefly summarize what is the bunch davies vacuum and how it can be selected by using the symmetries of the de sitter spacetime and the hadamard condition . further details can be found in refs . the action of the test massless scalar field , @xmath17 , is given by @xmath175 where @xmath176 denotes the determinant of the metric @xmath177 that in conformal time takes the form @xmath178 here , @xmath179 and @xmath180 with @xmath132 the constant hubble parameter ( in cosmic time ) . it is worth commenting that , due to homogeneity and isotropy of this spacetime , the metric is invariant under rotations and translations . in addition , it is invariant under the dilatations @xmath181 . these transformations leave @xmath132 unaltered . the consequence is that the previous action in eq . will be also invariant under this set of transformations . let us now consider the redefinition @xmath182 and decompose the field @xmath183 in fourier modes of the form of @xmath184 . then , the partial differential field equation can be written in terms of a set of infinitely many ordinary differential equations given by @xmath185 with @xmath186 . one can easily see that these equations are invariant under the set of transformations considered above , in agreement with the action in eq . . the solutions to these equations are known , @xmath187 where @xmath188 and @xmath189 are complex constants , that in principle may depend on the mode @xmath190 . let us remind that the choice of a particular vacuum for the fock quantization is tantamount to select a complete set of solutions to the equations of motion satisfying the normalization condition given in eq . . the normalization condition imposes @xmath191 and , taking into account the irrelevance of a complex global phase , the freedom on selecting a vacuum state is given by two real parameters per mode . nonetheless , if one imposes invariance of the vacuum under rotations then the complex constants can only depend on @xmath49 . in addition if one requires invariance under dilations then @xmath188 and @xmath192 must be k - independent , therefore @xmath193 and @xmath194 . the bunch davies vacuum is obtained by taking @xmath195 ( and therefore @xmath196 ) and it is the unique vacuum which is invariant under the afore mentioned symmetries and its 2-point function has the hadamard form @xcite . in order to elaborate more about this point , let us consider symmetry invariant vacuum states that we will denote as @xmath197 and obtain the 2-point function , for simplicity , evaluated at the same time @xmath46 . it is given by @xmath198 let us comment that the 2-point functions in this 2-parameter family are invariant under spatial translations and rotations , and dilations of the space and time . the 2-point functions of the field @xmath17 are easily related with the previous ones by @xmath199 one can compute explicitly these 2-point functions @xcite . if we expand the quantum field @xmath200 as @xmath201 such that @xmath202 , the expectation values in eq . can be computed and yield @xmath203 where @xmath204 stands for the imaginary part and the functions @xmath205 are defined as @xmath206 such that , @xmath207 here , we have introduced the function @xmath208 then , we immediately see that @xmath209 and @xmath210 oscillate as functions of @xmath46 . this property will be essential in our criterion for the choice of vacuum state . the previous integrals can be computed ( see ref . @xcite ) , and one can obtain explicitly @xmath211 . however , we will not give here the explicit result . nonetheless , one can check in ref . @xcite that among these 2-point functions , there is only one that is hadamard @xcite . precisely , it is the only 2-point function where the contributions of @xmath209 and @xmath210 to the power spectrum disappear since it corresponds to @xmath212 and @xmath213 . as we mentioned before , this choice corresponds to the so - called bunch - davies vacuum state . as we have shown any vacuum state is obtained from a complete set of complex solutions @xmath54 . therefore , given an initial time @xmath57 , the initial data @xmath113 that select a set of solutions for the bunch davies vacuum are @xmath214 taking into account that the addition of a global phase to the solutions still defines the same vacuum state , we will consider instead initial data @xmath215 of the form given in eq . with @xmath216 where the superscripts are referred to the bunch - davies vacuum . it is straightforward to see that with these initial data one gets the equivalent set of solutions @xmath217 , where @xmath218 . now , we will show the consequences of our criterion when it is applied to this particular quantum field theory . we will only assume invariance under translations and rotations in what follows . let us then start by choosing a set of arbitrary initial data @xmath219 accordingly , for a given initial time @xmath220 , of the form of eq . . one can easily see that the resulting set of solutions @xmath108 can be written as in eq . . this time , the coefficients , in principle , can depend on @xmath49 . indeed , if @xmath221 and @xmath222 , we get we will now concentrate only in the fourier transform of the two point function evaluated at the same time @xmath46 . it is essentially given by eq . as @xmath224 with @xmath225\frac{1}{2k}\left(1+\frac{1}{k^2\eta^2}\right),\ ] ] and@xmath226 - 2 \arctan\left[-\frac{1}{k \eta_0}\right].\ ] ] these are the basic ingredients we need in order to proceed with our method . we should minimize the integral in eq . with respect to @xmath227 and @xmath94 . numerically , it is possible to minimize this function for any choice of @xmath228 and @xmath229 . of course , for those modes that have not oscillated several times in the interval @xmath230 $ ] , this algorithm will not yield an accurate value of the minimum . this is why we must consider @xmath230 $ ] to be sufficiently extensive , covering a large portion of the spacetime . since it is computationally expensive , we will consider instead the minimization of the integral @xmath231 which yields the same result , at least for conformal time , but with this last expression easier to handle analytically than eq . . the integral can be computed explicitly . nevertheless , we will not show here the result since the expression is considerably long . indeed , it is enough to consider , for any fixed value of @xmath100 , the behavior of the integral when @xmath232 . it is possible to see that it diverges as @xmath233 for any @xmath49 if @xmath234 . here we have used @xmath235 . nevertheless , if @xmath236 for all @xmath49 , the integral always converges , since it behaves as @xmath237.\ ] ] we conclude that , for @xmath232 ( when all the modes are inside the hubble radius initially ) , there is only one minimum for this integral , and therefore , only one vacuum state , correspondingly . moreover , this vacuum state coincides with the already mentioned privileged de sitter vacuum that is invariant under spatial translations and rotations , dilations of space and time , and such that its 2-point function fulfills the hadamard condition ( i.e. the bunch - davies state ) . m. martn - benito , d. martn - de blas and g. a. mena marugn , _ approximation methods in loop quantum cosmology : from gowdy cosmologies to inhomogeneous models in friedmann walker geometries _ , class . quantum grav . * 31 * , 075022 ( 2014 ) . l. castell gomar , j. cortez , d. martn - de blas , g. a. mena marugn , j. m. velhinho , _ uniqueness of the fock quantization of scalar fields in spatially flat cosmological spacetimes _ , jcap 1211 , 001 ( 2012 ) . jeronimo cortez , daniel martin - de blas , guillermo a. mena marugan , jose m. velhinho , _ massless scalar field in de sitter spacetime : unitary quantum time evolution _ , class . quantum grav . 30 ( 2013 ) 075015 . b. bolliet , j. grain , c. stahl , l. linsefors , a. barrau , _ comparison of primordial tensor power spectra from the deformed algebra and dressed metric approaches in loop quantum cosmology _ , d * 91 * , 084035 ( 2015 ) .	we provide the power spectrum of small scalar perturbations propagating in an inflationary scenario within loop quantum cosmology . we consider the hybrid quantization approach applied to a friedmann robertson walker spacetime with flat spatial sections coupled to a massive scalar field . we study the quantum dynamics of scalar perturbations on an effective background within this hybrid approach . we consider in our study adiabatic states of different orders . for them , we find that the hybrid quantization is in good agreement with the predictions of the dressed metric approach . we also propose an initial vacuum state for the perturbations , and compute the primordial and the anisotropy power spectrum in order to qualitatively compare with the current observations of planck mission . we find that our vacuum state is in good agreement with them , showing a suppression of the power spectrum for large scale anisotropies . we compare with other choices already studied in the literature .
there are many important physical systems where the strong interaction at nonzero temperature and density plays a crucial role : e.g. , neutron stars , heavy - ion collision experiments , and the early universe . to understand these systems , we need to better grasp the phase diagram of qcd at nonzero temperature and chemical potentials . unfortunately , the most reliable techniques used at nonzero temperature and zero chemical potentials are generally not applicable at finite chemical potentials . in particular , lattice simulations suffer from the so - called sign problem . this problem has not yet been solved explicitly . however , some recent advances have allowed the study of the high temperature and low chemical potential part of the phase diagram . our current understanding of the phase diagram is therefore mainly based on various models which validity is difficult to test @xcite . in this article we shall use a hadron resonance gas model to study properties of the qcd phase diagram at nonzero temperature and chemical potentials . it is one of the models that has been successfully used in the past to study the hadronic phase at finite temperature and chemical potentials @xcite . this model is in good quantitative agreement with the lattice simulations for various observables in the hadronic phase @xcite . furthermore , it has also been used to determine the critical temperature that separates the hadronic phase from the quark - gluon plasma phase @xcite . we would like to stress here again that the agreement with the lattice simulations is excellent , especially at small chemical potential . recent lattice simulations at finite temperatures and zero chemical potentials seem to indicate that the critical temperature for the light quark - antiquark condensate is different than that for the strange quark - antiquark condensate @xcite . in particular , the lattice results for the different quark number susceptibilities indicate that the pseudocritical temperature of the crossover for the restoration of chiral symmetry for the light quarks is smaller than that for the strange quark @xcite . this difference has also been observed in chiral perturbation theory at zero chemical potential @xcite . in this article we shall use the hadron resonance gas model to study the possible differences between the behavior of the light @xmath0 and strange @xmath1 quark - antiquark condensates in the hadronic phase . at small chemical potentials , we find that the strange quark - antiquark condensate is still large where the light quark - antiquark condensates become very small . at higher chemical potentials , this difference slowly diminishes : the light and strange quark - antiquark condensates become small at the same temperature . the difference between the light and strange quark - antiquark condensates might lead to significant consequences for the heavy - ion collision experiments . one can expect the existence of a domain of temperatures where the chiral symmetry is restored for the light quark flavors and broken for the strange quark flavor . the pressure in the hadronic phase is given by the contributions of all the hadron resonances up to @xmath2 gev treated as a free gas @xcite . all thermodynamic quantities can be derived from the pressure @xmath3 where @xmath4 is the grand canonical partition function in a finite volume @xmath5 , at nonzero temperature , @xmath6 , baryon chemical potential , @xmath7 , isospin chemical potential , @xmath8 , and strangeness chemical potential , @xmath9 . it has been found that this model gives a very good description of the hadronic phase and of the critical temperature that separates the hadronic phase from the quark - gluon plasma phase @xcite . in the free gas approximation , the contribution to the pressure due to a particle of mass @xmath10 , baryon charge @xmath11 , isospin @xmath12 , strangeness @xmath13 , and degeneracy @xmath14 is given by @xmath15 where @xmath16 for fermions and @xmath17 for bosons . @xmath18 is the modified bessel function . in the hadronic phase the isospin is an almost exact symmetry since the hadronic spectrum is almost isospin symmetric @xcite . the quark - antiquark condensates are given by the derivative of the pressure with respect to the constituent quark masses : @xmath19 where @xmath0 represents the light quark - antiquark condensate . @xmath20 and @xmath21 indicate the value of the light and strange quark - antiquark condensates in the vacuum , respectively . the computation of the quark - antiquark condensates in eq . ( [ qqhrg ] ) requires modeling two quantities . the first one is the strange quark - antiquark condensate at zero temperature and zero chemical potential , @xmath21 . for it we need to know the kaon decay constant , which is unfortunately not yet known experimentally . but in the framework of lattice qcd , the ratio of heavy to light meson decay constants is found to be given by @xmath22 @xcite ( see also @xcite ) . using qcd sum rules , the ratio of strange to light quark - antiquark condensates in vacuum is given by @xmath23 @xcite ( another estimation gives @xmath24 @xcite ) . the gell - mann oakes renner relation connects the quark - antiquark condensates at zero temperature and zero chemical potential to the meson masses and to their decay constants . we use the next - to - leading order result obtained in chiral perturbation theory @xcite : @xmath25 where @xmath26 stands for the light quark mass where we assume @xmath27 . the coefficient @xmath28 has been obtained from the low - energy coupling constants of chiral perturbation theory @xcite . notice that @xmath29 does not contain any chiral logarithm . the second quantity which we have to model in order to calculate the quark - antiquark condensates in the hadron gas model eq . ( [ qqhrg ] ) is the quark mass dependence of the hadron masses . results from lattice simulations @xcite indicate that @xmath30 where @xmath31 . using these results together with the gell - mann oakes renner relation , eq . ( [ gor ] ) , one finds that @xcite @xmath32 relation ( [ mhms ] ) is not valid for the pion mass , since pions are almost independent of @xmath33 in chiral perturbation theory @xcite . however , relations ( [ mhmq ] ) and ( [ mhms ] ) work reasonably well for the nucleon @xcite , for instance . we therefrom assume that this relation is valid for all the hadrons heavier than the pions . this is in agreement with the lattice simulations , where the masses of a few hadrons have been shown to follow eq . ( [ mhmq ] ) and eq . ( [ mhms ] ) over a sizable range of quark masses @xcite . using these assumptions , we find that the contribution of one hadron of mass @xmath10 to the light quark - antiquark condensates is given by @xmath34 whereas its contribution to the strange quark - antiquark condensate is given by @xmath35 we include all hadron resonances with masses up to @xmath36 gev . the light and strange quark - antiquark condensates , eq . ( [ qq ] ) and eq . ( [ ss ] ) are calculated as a function of temperature @xmath6 and for various values of baryon , isospin , and strangeness chemical potentials , @xmath37 , @xmath8 , and @xmath38 , respectively . the results are shown in fig . [ fig1 ] and fig . [ fig2 ] . at zero chemical potential , we find that the strange quark - antiquark condensate is still large at temperatures where the light quark - antiquark condensates become small : @xmath39 where @xmath40 vanishes . as can be seen in fig . [ fig2 ] , an increase in @xmath7 tends to reduce the difference between @xmath40 and @xmath1 . increasing @xmath8 apparently accentuates this difference . we also notice that increasing @xmath9 does not sizably affect . the reason for this behavior is that the light quark - antiquark condensates in the hadron resonance gas model are much more sensitive to the pion physics than the strange quark - antiquark one . on the other hand , we find that all condensates are equally sensitive to @xmath7 . the value of @xmath41 at the critical temperature for the light quark flavors is shown in fig . [ fig3 ] as a function of @xmath7 . we study two cases : @xmath42 and @xmath43 and @xmath42 and @xmath44 . the latter is very much compatible with the corresponding values calculated for recent heavy - ion collision experiments , as mentioned above . we conclude that the strange quark - antiquark condensate can be relatively large at temperatures where the light quark - antiquark condensate is very small . we obtained these results from a hadron resonance gas model . our results agree with the recent milc lattice simulations at zero temperature and chemical potentials @xcite . the uncertainty of the value of the strange quark - antiquark condensate at this critical temperature is however rather large . this is due to the uncertainty in the values of the condensates at high temperatures which we have calculated by the hadron resonance gas model . this uncertainty could be reduced for the most part , if the factor @xmath45 in eq . ( [ lattmass ] ) were more precisely known . we have shown that in the hadron resonance gas model , the strange quark - antiquark condensate can still be sizably large at temperatures where the light quark condensates nearly vanish . in other words , the hadron resonance gas model seems to indicate that the chiral symmetry could still broken for the strange quark flavor at temperatures where chiral symmetry is restored for the light quark flavors . this new situation seems to be possible only in a range of a few tens of mev around @xmath46 mev . this result is in agreement with recent lattice calculation at zero chemical potential . we have shown that this effect persists for small chemical potentials and that it seems to slowly disappear at higher chemical potentials . we have also shown that this effect can be enhanced by a nonzero isospin chemical potential . this fact could be used in heavy - ion collision experiments by choosing different isotopes , which is an interesting option for other reasons as well @xcite . this effect does not contradict any fundamental principle . at nonzero quark masses , lattice simulations show that in qcd the hadronic phase and the quark - gluon plasma phase are separated by a crossover , not by a sharp phase transition . it is therefore possible for the light quark - antiquark condensates to be small while the strange quark - antiquark condensate is large in the crossover region . however , it is generally expected that this crossover stops and becomes a first order phase transition with increasing @xmath7 . in this latter case , @xmath40 and @xmath1 should drop and become small at the same temperature . furthermore , the discontinuity of @xmath40 and @xmath1 at a first order phase transition are related by a clausius - clapeyron relation : they should be similar @xcite . the emergence of a first order phase transition at nonzero @xmath7 is therefore entirely possible in the hadron resonance gas model , since the difference between @xmath40 and @xmath1 is smaller at larger @xmath7 . notice however , that the difference between @xmath40 and @xmath1 is larger for larger @xmath8 . this would tend to indicate that there is no critical endpoint in the qcd phase diagram at nonzero temperature and isospin chemical potential . finally , we observe that if this effect is true , our study opens the possibility of the existence of a domain where chiral symmetry is partially restored with a spontaneous chiral symmetry breaking given by @xmath47 . it would be interesting to study the physics of such a spontaneous symmetry breaking of chiral symmetry and determine its consequences for heavy - ion collision experiments . z. fodor and s. d. katz , phys . b * 534 * , 87 ( 2002 ) ; jhep * 0203 * , 014 ( 2002 ) ; jhep * 0404 * , 050 ( 2004 ) ; c. r. allton _ et al . _ , phys . d * 66 * , 074507 ( 2002 ) ; phys . d * 68 * , 014507 ( 2003 ) ; f. karsch _ et al . _ , nucl . proc . suppl . * 129 * , 614 ( 2004 ) ; c. r. allton _ et al . _ , phys . d * 71 * , 054508 ( 2005 ) . p. de forcrand and o. philipsen , nucl . b * 642 * , 290 ( 2002 ) ; nucl . phys . b * 673 * , 170 ( 2003 ) ; nucl . suppl . * 129 * , 521 ( 2004 ) . m. delia and m. p. lombardo , phys . d * 67 * , 014505 ( 2003 ) . r. v. gavai and s. gupta , phys . d * 68 * , 034506 ( 2003 ) ; r. gavai , s. gupta and r. roy , prog . . phys . suppl . * 153 * , 270 ( 2004 ) . v. azcoiti , g. di carlo , a. galante and v. laliena , arxiv : hep - lat/0503010 . for a review see , k. rajagopal and f. wilczek , arxiv : hep - ph/0011333 ; for a review see , p. braun - munzinger , k. redlich and j. stachel , arxiv : nucl - th/0304013 . f. karsch , k. redlich and a. tawfik , eur . j. c * 29 * , 549 ( 2003 ) ; phys . lett . b * 571 * , 67 ( 2003 ) ; f. karsch , prog . suppl . * 153 * , 106 ( 2004 ) ; k. redlich , f. karsch and a. tawfik , j. phys . g * 30 * , s1271 ( 2004 ) , m. delia and m. p. lombardo , arxiv : hep - lat/0406012 ; a. tawfik , hep - ph/0410329 , a. tawfik , phys . d * 71 * , 054502 ( 2005 ) ; d. toublan and j. b. kogut , phys . b * 605 * , 129 ( 2005 ) . c. w. bernard , nucl . phys . proc . suppl . * 94 * , 159 ( 2001 ) [ arxiv : hep - lat/0011064 ] . h. leutwyler and m. roos , z. phys . c * 25 * , 91 ( 1984 ) . m. jamin , phys . b * 538 * , 71 ( 2002 ) [ arxiv : hep - ph/0201174 ] . s. narison , arxiv : hep - ph/0202200 . j. gasser and h. leutwyler , nucl . b * 250 * , 465 ( 1985 ) . m. e. sainio , pin newslett . * 16 * , 138 ( 2002 ) . b. klein , d. toublan and j. j. m. verbaarschot , phys . d * 68 * , 014009 ( 2003 ) ; d. toublan and j. b. kogut , phys . b * 564 * , 212 ( 2003 ) ; m. frank , m. buballa and m. oertel , phys . b * 562 * , 221 ( 2003 ) ; a. barducci , r. casalbuoni , g. pettini and l. ravagli , phys . d * 69 * , 096004 ( 2004 ) . a. barducci , g. pettini , l. ravagli and r. casalbuoni , phys . b * 564 * , 217 ( 2003 ) .	we use a hadron resonance gas model to calculate the quark - antiquark condensates for light ( up and down ) and strange quark flavors at finite temperatures and chemical potentials . at zero chemical potentials , we find that at the temperature where the light quark - antiquark condensates entirely vanish the strange quark - antiquark condensate still keeps a relatively large fraction of its value in the vacuum . this is in agreement with results obtained in lattice simulations and in chiral perturbation theory at finite temperature and zero chemical potentials . furthermore , we find that this effect slowly disappears at larger baryon chemical potential . these results might have significant consequences for our understanding of qcd at finite temperatures and chemical potentials . concretely , our results imply that there might be a domain of temperatures where chiral symmetry is restored for light quarks , but still broken for strange quark that persists at small chemical potentials . this might have practical consequences for heavy ion collision experiments .
solar corona , a tenuous and very hot outer part of the solar atmosphere which sits on top of other layers ( photosphere , chromosphere , and the transition region ) can be seen either in white light observations during the solar eclipses ( because of @xmath0 times more flux of visible photospheric photons is otherwise outshining it ) or in extreme ultra - violet and x - ray observations from space ( because of atmospheric absorption of the short wavelength radiation ) . @xcite describe the first successful observations of the green coronal line @xmath1 by w. harkness and independently by c.a . young during the total solar eclipse of august 7 , 1869 . this `` new '' line was attributed to a new chemical element `` coronium '' . it was easy to make such a mistake because just a year before , 1868 , during the solar eclipse @xcite discovered a prominent yellow line @xmath2 , which could not be ascribed to any known chemical element at the time . this marked a discovery of helium after greek word helios meaning sun. on earth helium was only found about 10 years later by w. ramsay . the blunder with coronium went on for quite some time . moreover , as described by @xcite , the new gas geo - coronium extends from 200 km above earth into the space , and that the steady auroral arcs are due to electric luminescence in this gas . the truth had to wait until the beginning of 1940s . in his george darwin lecture of the royal astronomical society @xcite entitled `` the identification of the coronal lines '' edln describes the process of realising that the mysterious coronal lines , including the green coronal line @xmath3 , come from so - called forbidden lines of highly ionised metals . for example , the most intense green coronal line comes from fexiv , iron atom with 13 electrons stripped off . edln rightfully acknowledges that the correct identification of the coronal lines was triggered by a letter of 13 february 1937 sent to him by w. grotrian . these forbidden lines are emitted from energetically unfavourable metastable atomic energy levels . under normal laboratory conditions electrons do not readily make transitions from the metastable states to lower energy states ( whilst emitting photons ) due collisional de - excitation . in the extremely rarefied conditions of the solar corona such transitions become possible . the highly ionised metal ions can only be created in extremely hot conditions @xmath4 few @xmath0 k. thus , at the time it was rather difficult to accept that solar corona is so hot . especially because this seemed in an apparent violation of the second law of thermodynamics ( in rudolf clausius s formulation ) that the heat can not flow from a cold body ( photosphere at @xmath5 k ) to a hotter body ( corona at few @xmath0 k ) . in this manner , one of the major unsolved problem in solar physics has emerged : which physical mechanisms are responsible for heating of solar corona ? some significant progress has been made , partcitlucalry , with the advent of space era in solving the coronal heating problem ( see e.g. chapter 9 in @xcite ) . as it happens in science , in general , the progress has led to yet more , previously unknown challenges such as finding the answers to the questions : ( i ) why solar wind speed in the polar regions is twice higher during the solar minimum than plausible solar wind models can explain ? ( ii ) why the solar flares , the violent events in the corona when magnetic energy is converted into other forms , occur on timescales much shorter than the plausible resistive timescales predict ? ( iii ) why the most of the solar flare energy is taken away by the accelerated , super - thermal particles ? and so on . it is important to realise that reproducing or mimicking the reality in ( mostly ) numerical models does not necessarily mean that we have an _ understanding _ of physical mechanisms in action . here are a few examples e.g. in simple 1d coronal loop models a steady state of few @xmath0 k plasma can be maintained by balancing an _ ad hoc _ heat input with realistic losses such as heat conduction , radiation , etc , or even time - varying flare dynamics can be simulated @xcite . more realistic 3d models @xcite mimic the solar coronal observations and yet use unrealistically high magnetic resistivity . no wonder if one puts sufficient heating or artificially adjusts dissipation coefficients , a few @xmath0 k plasma can be obtained that `` looks like '' the solar corona . but did we learn what generates this heat or makes the dissipation coefficients anomalously high ? albeit , the answer is no . notwithstanding , something useful can be learnt e.g. where spatially heat needs to be deposited to reproduce the observations . similarly , consider the sophisticated solar wind models which solve numerically fluid - like equations for electrons , protons and alpha particles @xcite . indeed , such models can reproduce twice as high solar wind speeds in the polar regions than near the solar equatorial plane , as observed during solar minima . but , this is obtained based on an _ ad hoc _ energy flux injected into the both ion species . again , it comes as no surprise that _ ad hoc _ energy flux ( essentially an additional momentum added ) to the ion species produces observed fast solar wind speeds . again , no answer is provided to the main question : what provides this additional energy flux ( momentum ) ? is it due to absorption of waves ? or due to some microscopic reconnection events at the base of the corona ? as to the question of why the most of the solar flare energy is taken away by the accelerated particles , the situation is analogous . kinetic - scale particle - in - cell modelling @xcite can in principle reproduce the solar flare observations in that within one alfvn time , somewhat less than half ( @xmath6% ) of the initial total ( roughly magnetic ) energy is converted into the kinetic energy of electrons , and somewhat more than half ( @xmath7% ) into kinetic energy of ions ( similar to solar flare observations ) . also , a sizable fraction ( up to 20% ) of the magnetic energy can be released / conversted into other forms such as kinetic energy of super - thermal particles and to much small extent waves . however , such simulations are performed usually in a domain sizes of several hundreds of electron debye length @xmath8 m , which is about @xmath9 times smaller than the solar flare particle acceleration site . here @xmath10 is electron thermal speed taken at @xmath11k and @xmath12 is electron plasma frequency taken at @xmath13 m@xmath14 , commensurate to solar corona . the question is whether realistic behaviour mimicked in small simulation domains will also hold if they are up - scaled to realistic sizes ? the situation is analogous to the stability of plasma in thermonuclear fusion devices such as tokamaks : small - scale ( `` laboratory''-scale ) versions behave quite differently from the plasma stability point of view than the large , `` factory''-scale ones . the above discussion about the spatial scales , @xmath15 , the characteristic length scales of a physical system over which properties of physical quantities ( e.g. magnetic field of an active region in the solar corona ) change brings us to the following dichotomy or a dilemma . one one hand , solar corona is well described by the mhd approach . this is because if we take @xmath15 as the hydrostatic scale - height of the corona ( height over which pressure drops @xmath16-times ) @xmath17 m , this is much larger than all relevant kinetic scales : ( i ) @xmath18 m , ion ( proton ) inertial length ( also called ion skin depth ) , ( ii ) @xmath19 m electron inertial length ( also called electron skin depth ) , ( iii ) @xmath20 m electron larmor radius . ( @xmath21 is the electron cyclotron frequency , which quantifies rotation of electrons around the magnetic field on their helical path . ) on the other hand , all interesting physical processes such as wave dissipation , nano - scale _ fast _ reconnection ( coronal plasma heating ) and particle acceleration ( during flares ) occur at small , kinetic scales . the dichotomy , depicted in fig . 1 , is in that the bulk plasma dynamics in the solar corona ( where most of the energy is stored ) is observable remotely from earth ( typical smallest coronal structures resolved e.g. with trace satellite is 0.5@xmath22 i.e. @xmath23 m on the sun ) . but the above interesting physical processes all operate at scales below 10 m. thus , unless an in - situ probes are sent to the solar corona the kinetic - scales processes will never be observed . it seems unlikely humankind will ever come up with materials ( for the probes ) which will withstand the heat of the coronal environment . however , esa has plans for proba mission ( @xmath24 ) , in which the telescope of proba-3 s solar coronagraph will be mounted on one of the spacecraft , while the other craft is manoeuvred to accurately occult ( mask ) the main disc of the sun . with the proposed proba-3 arrangement , it is anticipated that accurate measurements will be possible from 1.05 to 3.2 solar radii . note that the discussion about spatial scales @xmath15 can always be cast into temporal scales , @xmath25 or frequencies @xmath26 by using characteristic speed of processes @xmath27 via @xmath28 . for example @xmath27 can be the alfvn speed , @xmath29 , if mhd - scale processes are considered . this would equally be applicable to waves and reconnection . in the latter case reconnection outflow speeds would be implied , while in the case of former phase speed of the wave . , ( i.e. @xmath30 ) , versus @xmath31 ) of magnetic fluctuations in the solar wind . adopted from @xcite . ] let us briefly discuss conditions of when the above mentioned kinetic scales become important . strictly speaking mhd equations are scale - free , in a sense that there is no physical spatial scale appearing in the equations ( except @xmath15 which is an _ arbitrary _ size of the system , which simply needs to stay macroscopic so that continuum mechanical ( fluid - like ) description remains valid ) . the devil is in the detail , however . in order to obtain closed set of equations one needs to express the electric field by means of other physical quantities . in mhd , the induction equation for the magnetic field is obtained by plugging in @xmath32 into the faraday s law ( from the maxwell s equations ) @xmath33 . intrinsic physical scales start appearing when instead of usual mhd - version generalised ohm s law @xcite , p. 89 , @xmath34 one starts to use a more general form : @xmath34 @xmath35 eq.(2 ) essentially can be obtained from the electron equation of motion if electron speed @xmath36 is replaced by @xmath37 . the latter is mathematical statement of : when electron and ion speeds are different , this causes charge separation electric field and hence current @xmath38 . in mhd , however , ion and electron separate dynamics is ignored and @xmath39 . it can be shown @xcite that each of the non - mhd terms ( second line of eq.(2 ) ) have intrinsic spatial scales associated with them : ( i ) the term containing @xmath40 ( called hall term ) becomes important on scales comparable to @xmath41 , ion skin depth ; ( ii ) the term containing @xmath42 ( called electron inertia term ) becomes important on , @xmath43 , electron skin depth scale ; and finally ( iii ) the term containing the divergence of the pressure tensor , @xmath44 , ( called pressure tensor term ) becomes important on scales comparable to @xmath45 , electron larmor radius . in order to better understand ordering of different spatial scales let us consider a simple problem of steady reconnection when oppositely directed magnetic field lines are brought into a diffusion region by an inflow of plasma , then they change connectivity and plasma outflow carries the field lines away . the change of connectivity occurs in the diffusion region which has width @xmath46 and length @xmath47 . in mhd , in the simplest possible formulation ( sweet - parker model ) : ( i ) bernoulli equation determines the outflow speed being the alfvn speed , @xmath48 ; ( ii ) continuity equation prescribes in the reconnection inflow speed @xmath49 ; while ( iii ) the generalised ohm s law in which the advection term @xmath50 is balanced by the resistive term @xmath51 sets the reconnection rate @xmath52 , where @xmath53 is the magnetic lundquist number . thus , in mhd sweet - parker model ( in which @xmath47 is fixed at @xmath15 ) is producing reconnection rates much smaller what is observed in e.g. flares in solar corona . for the parameters commensurate to solar corona ( @xmath54 m@xmath14 , @xmath55 k , @xmath56 m , @xmath57 t ( 100 gauss and hence @xmath58 m s@xmath59 ) , @xmath60 ) the classical sweet - parker rate is @xmath61 . the reconnection rate is also interpreted as the ratio of alfvn time ( @xmath62 s ) and resistive ( or reconnection ) times . this means in the sweet - parker model resistive ( or reconnection ) time is @xmath63 s @xmath64 days . on contrary , in the observations flares last for up to tens of minutes . petschek model alleviates this problem by making the diffusion region length , @xmath65 . this means that the plasma inflow speed @xmath49 a sizable fraction of the alfvn speed making reconnection rates faster and hence flare times shorter commensurate to the observations . however , petschek model is a phenomenological one ( not derived from the first principles ) and for it to work ( as demonstrated by numerous numerical simulations ) it requires plasma resistivity to be non - uniform and be enhanced on the diffusion region ( so - called anomalous resistivity ) . whilst , the latter being quite plausible , given that turbulent transport is known to enhance the dissipation / resistive coefficients dramatically , we have not sent in - situ probes to the solar corona yet , and hence we do not know for certain whether it is turbulent or not ! it should be noted that in the laboratory plasma magnetic reconnection experiment ( mrx ) measurements show indication of the reconnection rates or equivalently anomalous resistivity values 10s and even 100 times larger that sweet - parker model predicts @xcite . the mrx experiment plasma _ is not in the turbulent _ state . what is observed , however , is that the reconnection rates or equivalently anomalous resistivity values are attained when the transition from collisional ( resistive mhd ) to collisionless ( kinetic ) regime takes place . ) and below the magnetic field is carried by electrons . black box with width @xmath43 shows electron diffusion region where magnetic connectivity change take place . the green box with width @xmath45 shows region where electron meandering motion makes pressure to become a tensor with off - diagonal ( non - gyrotropic ) terms playing a major role . blue and black dashed lines with arrow heads show ion and electron flow , respectively . note how ion flow deflects at the scale @xmath41 , while electron flow does so at @xmath43 . the green wiggle in the centre show an electron orbit . its radius is small and follows a magnetic field line away from the diffusion region . the radius becomes large and motion chaotic ( meandering ) as the magnetic field drops to zero close to the magnetic null . ] so , is there an alternative way ( i ) to have fast reconnection i.e. get solar flare times right ? ( ii ) also to demonstrate that the large proportion of flare energy goes into the accelerated particles ? and ( iii ) not to make assumptions about the turbulent state of the corona ( hence postulating the anomalous resistivity ) ? it turns out that plasma _ kinetic _ approach can provide answer `` cautious yes '' to all of these fundamental questions . here is how : for the solar coronal parameters : a temperature of @xmath66 k , coulomb logarithm of 18.1 , the lundquist number ( using spitzer resistivity ) is @xmath67 . here @xmath56 m was used . one of the reasons for going beyond resistive mhd is comparing typical width of a sweet - parker current sheet @xmath68 m to the ion skin depth . typical scale associated with the hall term in the generalised ohm s law at which deviation from electron - ion coupled dynamics is observed is , @xmath69 m. here particle number density of @xmath70 m@xmath14 has been used . hence , the fact that @xmath71 justifies going _ beyond single fluid resistive mhd approximation _ ( a similar conclusion is reached by @xcite in their fig . in other words , ion skin depth scale is reached first before resistive mhd sweet - parker current sheet . let us put this fact ( that @xmath72 ) in the visual context using fig . 2 : on a global ( bulk mhd ) scale @xmath15 electrons and ions inflow towards the diffusion region ( where magnetic field connectivity changes ) in a coordinated way as if they are glued to each other . when the flow approaches ion skin depth scale @xmath41 ions become demagnetised ( magnetic field is no longer is frozen into ion fluid ) and magnetic field is carried forward by electrons . at @xmath41 scale the hall term in the generalised ohm s law dominates . this picture is broadly corroborated by kinetic , particle - in - cell simulations of x - point collapse @xcite . in panel ( a ) of fig.3 we see that for electrons the diffusion region width is indeed @xmath43 , while for ions , panel ( b ) of fig.3 , the width of the region where ions start to deflect from the current sheet ( region where ions become demagnetised ) is about @xmath73 ( note that here the ion to electron mass ratio of @xmath74 has been used ) . this continues until electron skin depth scale @xmath43 is reached where electron inertia term takes over . further down the spatial scales , when electron larmor radius scale @xmath45 is reached electron meandering motion ( dashed green wiggled curve in fig.2 ) sets in @xcite . this makes the non - gyrotropic ( off - diagonal ) components of the electron pressure tensor to dominate other terms in the generalised ohm s law , and hence makes them responsible for breaking the frozen - in condition . the concept of pressure _ tensor _ is somewhat counter - intuitive . this can be comprehended as follows : away from the electron diffusion region pressure is scalar ( isotropic ) and electrons flow along ( are frozen into ) the magnetic field lines . inside the the diffusion region @xmath75 electrons become demagnetised and execute meandering ( rather chaotic ) motions . these are such that pressure becomes different as one travels in different directions ( i.e. pressure becomes a tensor as opposed to a scalar ) . previous results on collisionless reconnection both in tearing unstable harris current sheet @xcite and stressed x - point collapse @xcite have shown that magnetic field is frozen into electron fluid and the term in the generalised ohm s law that is responsible for breaking the frozen - in condition is electron pressure tensor off - diagonal ( non - gyrotropic ) component gradients . thus , there is a case for inclusion of the _ electron pressure tensor non - gyrotropic components _ in a model of collisionless reconnection . analytically , this was achieved by @xcite . in this simple model instead of balancing the advection term in the generalised ohm s law with the resistive term ( as discussed above ) , non - gyrotropic components of the electron pressure tensor gradient were used . this produced simple analytical results which explain well ( i ) the reconnection rate and ( ii ) width of the electron diffusion region measurements in mrx experiment . , while ion flow deflects from the diffusion region at a width @xmath41 ( here @xmath76 ) , as sketched in fig . 2.,title="fig : " ] , while ion flow deflects from the diffusion region at a width @xmath41 ( here @xmath76 ) , as sketched in fig . 2.,title="fig : " ] there is a growing amount of work @xcite that suggests that in the collisionless regime , on the scales less than @xmath41 magnetic field is frozen into the electron fluid rather than bulk of plasma . one can write in general @xmath77 . this relation clearly shows that in collisional regime ( when the number density @xmath78 is large ) , the difference between electron and ions speeds diminishes @xmath79 . however , as one enters collisionless regime ( when the number density @xmath78 is small ) the deviation between electron and ion speeds starts to show . in @xcite we proposed a possible explanation why the reconnection is fast when the hall term is included . inclusion of the latter means that in the reconnection inflow magnetic field is frozen into _ electron _ fluid . as it was previously shown in @xcite ( see their figs.(7 ) and ( 11 ) ) speed of electrons , during the reconnection peak time , is at least 4 - 5 times greater than that of ions . this means that electrons can bring in / take out the magnetic field attached to them into / away from the diffusion region much faster than in the case of single fluid mhd which does not distinguish between electron - ion dynamics . thus , it is clear that inclusion _ magnetic field transport by electrons _ is crucial in resolving of above mentioned problems mhd description of the solar corona faces . to summarise , some of the time - dependent collisionless reconnection models such as tearing unstable harris current sheet @xcite and stressed x - point collapse @xcite where able : ( i ) to show that a sizable fraction of the magnetic energy can be converted / released into heat and accelerated , super - thermal particle energy ( see e.g. fig.13 from @xcite ) ; ( ii ) obtain fast reconnection rates and hence magnetic energy release rate commensurate to solar flare observations ( see e.g. figs.3 and 9 from @xcite ) ; ( iii ) reproduce the observational fact that the large proportion of flare energy goes into the accelerated particles in the correct partition , i.e. somewhat less than half ( 40% ) of the initial total ( roughly magnetic ) energy is converted into the kinetic energy of electrons , and somewhat more than half ( 60% ) into kinetic energy of ions ( see e.g. figs.3 and 4 from @xcite ) ; ( iv ) not to invoke anomalous resistivity . however , despite this success there are main challenges ahead : ( i ) the processes described by the collisionless reconnection models are small scale ( @xmath80 m ) . indeed , on scale less than ion skin depth magnetic field is advected into the diffusion region by electrons and since electrons are lighter and move faster , the reconnection is accordingly fast . but , sufficiently far away from the electron diffusion region , electron flow must slow down to bulk plasma speeds when mhd description takes over ( ions and electrons become glued to each other ) , where and how this happens ? recent pic simulations @xcite suggest that the electron diffusion region length can extend for much longer distances downstream than previously thought . why does this happen is not understood . ( ii ) how this small - scale dynamics on the kinetic scale translate into mhd scales ? i.e. despite producing fast reconnection rates , these processes happen in physically small volumes . will they affect mhd - scale ( say a @xmath81 m ) processes ? it will be impossible to find out until we can perform a numerical simulation @xmath0m@xmath82m@xmath82 m using realistic coronal active region fields as an input for particle - in - cell or vlasov simulation . the latter would seem many decades away even if moore s law ( that the number of transistors that can be placed inexpensively on an integrated circuit has increased exponentially , doubling approximately every two years ) continues to hold in the future . until such times , we can do @xmath83m@xmath84m@xmath84 m kinetic simulation and hope for the best that our models will scale correctly towards the larger ( physically meaningful ) spatial scales . study of mhd waves in the solar corona has seen steady progress over last two decades . there are two aspects to the study : ( i ) waves carry information about the medium they propagate in . this is used in the field of solar coronal seismology to infer physical parameters of the corona ( see for reviews @xcite ) . for example traditional method of measuring magnetic field in solar photosphere is based on the zeeman effect . however , in solar corona plasma is too dilute rendering this method unusable . based on the seminal theoretical work of @xcite , transverse oscillations of solar coronal euv loops were used to successfully measure the magnetic field @xcite . ( ii ) waves transport mechanical ( and magnetic ) energy and if dissipated may produce plasma heating . in the mhd approximation , in a uniform plasma penetrated by a uniform magnetic field , @xmath85 , there are three distinct types of waves : fast and slow magnetosonic and alfvn waves . fast magnetosonic waves are compressible mhd waves which have an advantage of propagating across the magnetic field . this implies that the heat produced by the wave dissipation can spread across and cover a large area . the fast magnetosonic waves that are generated in the solar interior can not reach the corona because of the reflection on the pressure / density gradient ( due to gravitational stratification ) . thus , initially they were excluded from coronal heating models . however , in the transversely inhomogeneous plasma there are two possibilities to couple alfvn waves ( which have no problem of crossing the gradient ) with fast magnetosonic waves : either by non - linear coupling in 2.5 mhd approximation @xcite ; or linear ( much more efficient ) coupling in 3d mhd approximation @xcite . slow magnetosonic waves are also compressible , but propagate preferentially along the magnetic field . normally , these waves are discounted from coronal heating candidate list on the grounds of not carrying ( less by four orders of magnitude ) sufficient flux @xcite . however , the latter estimate is based on the assumption of just a single harmonic ( fixed frequency ) and @xcite have demonstrated that if a wide spectrum ( continuous spectrum with a plausible frequency ( or wavenumber ) range ) of slow magnetosonic waves exists in the corona , than their dissipation would provide sufficient heating . whether such spectrum exists is an open question . however since in the nature there is always a cascade of energy from large scales to small scales ( see e.g. fig . 1 ) , this wide spectrum conjecture seems plausible . alfvn wave is an incompressible , transverse wave ( such as e.g. electromagnetic wave ) , with frequency , @xmath86 , much smaller that ion cyclotron frequency , @xmath87 . in an alfvn wave the background magnetic field tension provides restoring force , while plasma ions provide inertia ( the two key ingredients for any oscillatory motion ) . alfvn waves are easy to excite , e.g. a sheared plasma flow across magnetic field . however , they are notoriously difficult to dissipate because of small resistivity of the solar corona . i.e. alfvn wave is a good vehicle for transporting energy ( e.g. from convective motions below the photosphere into the corona ) , but this `` vehicle '' has virtually no breaks ( hence can not be stopped ! ) . thus , in order to deposit energy in the first @xmath17 m of the corona ( the hydrostatic scale - height ) above the transition region , some enhanced dissipation mechanisms where proposed in the past . in the solar corona , basic magnetic structure that is encountered is either a closed loop ( during 11-year cycle minimum more of these are found near solar equator and low latitudes ) ; or open magnetic strictures such a plumes ( during the solar minimum these are mostly located near the north and south polar regions ) . the common feature of these coronal structures is the density variation _ across _ the magnetic field . i.e. inside the loops and plumes density is enhanced by 3 10 times compared to the surrounding plasma . in the case of closed magnetic structures this enhanced density is believed to be due to material evaporation from denser layers of the sun such as transition region and chromosphere . in the open magnetic structures the density enhancement is either due to a solar wind and again material evaporation from below ( which could be either due to reconnection or any other process which increases plasma temperature so that the balance no longer holds ) . there are two mechanisms for the above mentioned enhanced alfvn wave dissipation and absorption in the solar corona : phase mixing @xcite and resonant absorption @xcite . in both mechanisms transverse inhomogeneity of plasma ( i.e. plasma with density variation across the uniform magnetic field ) plays a crucial role ( albeit for different reasons ) . brief review of both topics can be found in @xcite . for particle acceleration and efficient _ non - resonant _ wave - particle interactions ( e.g. converting wave energy to particle kinetic energy and heat if inter - particle collisions are efficient ) , it is necessary for a wave in question to have magnetic field - aligned electric field . the word non - resonant is crucial here because , interaction of waves with particles when ( i ) the wave phase speed @xmath88 coincides with particle ( thermal ) speed @xmath89 leads to the landau resonance ; or ( ii ) to the cyclotron resonance , when the rotation frequency of the electric field vector of the circularly polarised wave , @xmath90 coincides with the particle s cyclotron frequency , @xmath91 ( and when both the wave electric field vector and particle rotate in the same direction ) . in the both cases particle experiences almost constant electric field , which facilitates efficient wave - particle resonant interaction . away from the resonance @xmath92 or @xmath93 in the case of landau damping ; or @xmath94 or @xmath95 in the case of cyclotron resonance , wave electric field oscillates too fast or too slow and has no effect on plasma particles of species @xmath96 . in the mhd approximation alfvn wave propagates strictly along the magnetic field ( with some finite parallel to the magnetic field wave number @xmath97 ) as shown in case a , fig . 4 . in this case the perpendicular wave number @xmath98 . for the field aligned electric field to exist @xmath99 condition should be true . case c , fig.4 also shows why the fast magnetosonic waves are believed to be efficient particle accelerators . this because they preferentially propagate across the magnetic field and hence posses finite @xmath99 . for alfvnic waves @xmath99 ( see case b , fig . 6 ) only possible when kinetic - scale effects are included . in particular , it has been realised @xcite that when externally applied wave frequency becomes resonant with the alfvn frequency @xmath100 , a resonance occurs and the driving wave converts into kinetic alfvn wave ( kaw ) which has the finite field - aligned electric field . more generally , generation of the parallel ( magnetic field - aligned ) electric field occurs when the applied low - frequency ( @xmath101 ) alfvn wave has @xmath102 comparable to any of the _ kinetic spatial scales_. such alfvn waves ( with @xmath103 ) are called _ dispersive alfvn waves _ ( daw ) @xcite . the latter split into two classes : kinetic alfvn waves ( kaw ) and inertial alfvn waves ( iaw ) . the distinction between the two is drawn according to the value of plasma beta , which is the ratio of thermal and magnetic pressures , i.e. @xmath104 . plasma beta can also be related to the ratio of the sound speed and alfvn speed squared , i.e. @xmath105 , the ratio of specific heats @xmath106 in e.g. adiabatic case is 5/3 . here @xmath107 is the sound speed . physically , plasma beta prescribes importance of either thermal effects or magnetic field . e.g in solar corona @xmath108 meaning that magnetic fields play a dominant role in coronal plasma behaviour . in the context of dispersive alfvn waves plasma beta prescribes which term in the generalised ohm s law supports parallel electric field @xcite . if @xmath109 , i.e. @xmath110 then the dominant mechanism for generation of parallel electric field in the generalised ohm s law is the electron inertia and one has inertial alfvn waves ( iaw ) . if @xmath111 , i.e. @xmath112 then the dominant mechanism for generation of parallel electric field in the generalised ohm s law is electron pressure gradient and one has kinetic alfvn waves ( kaw ) . in the solar corona , particle acceleration e.g. during solar flares is generally believed to be occurring as a result of magnetic reconnection . for description of particle dynamics either test particle approach @xcite which is ignores self consistent fields or more recently kinetic , particle - in - cell simulation @xcite has been used . the problem with particle acceleration during solar flares is associated with the numbers of particles involved . in the standard flare model , reconnection event occurs higher up the corona and as result 50 - 80 % of the flare energy is converted into accelerated particles ( electrons and ions ) . these rush down towards denser layers of the solar atmosphere guided by coronal magnetic field structures and produce x - ray emission via bremsstrahlung . the number of detected x - ray photons can be related ( via certain model assumptions ) to the number of accelerated electrons . for a typical flare more than @xmath113 electrons are accelerated , yielding a huge current sheet ( with a volume of @xmath114 @xmath115 ) that must remain stable for more than 60 seconds . this is an unlikely course of events because of the dissolution ( thinning ) of the current sheet @xcite . in effect , the above numbers imply that all of the electrons in the current sheet need to be evacuated . another problem is with the return currents . a beam of electrons injected into plasma is known to be compensated by the generated counter beam . thus , expecting a bunch of all @xmath113 electrons happily travels towards the footpoints of the magnetic structure is naive . a viable alternative for transporting flare energy by the beam of accelerated particles , is waves . as we know waves can transport energy without mass transport . therefore , if flare energy is delivered to footpoints by waves then the above mentioned difficulties ( the number problem and return currents can be avoided ) . in this scenario , reconnection flare event produces dispersive alfvn waves ( which posses parallel electric fields ) . these subsequently travel to the footpoints , then accelerate particles in the vicinity of footpoints . the accelerated particles ( mostly electrons , because these waves are known to preferentially accelerate electrons due to their small inertia , whilst producing ion heating broadening of their distribution function ) , in turn , produce the observed x - rays via the bremsstrahlung . using particle - in - cell simulation , @xcite have recently explored how circularly polarised alfvn wave with @xmath116 propagates in a transversely inhomogeneous plasma . this frequency is somewhat higher than that for mhd wave and yet smaller that ion cyclotron resonance . such wave strictly speaking is called ion cyclotron rather than alfvn wave . in the @xmath117 limit , however , is behaves like an alfvn wave . the density in the middle of simulation domain was smoothly increased by a factor of 4 ( a ramp - like function @xmath118 $ ] ) across the uniform magnetic field in order to model a coronal loop ( accordingly temperature was varied as inverse of density so that total pressure balance is preserved ) . the widths of the transverse inhomogeneity on each side of the ramp is @xmath119 . fig . 5 s left column demonstrates that initially plane alfvn wave front becomes distorted ( middle part travels slower ) because its phase speed ( @xmath120 ) is a function of density , which in turn is a function of a coordinate @xmath121 across the uniform magnetic field applied along @xmath122-axis . it is also interesting to note that in the strongest density gradient regions @xmath123 and 150 the propagating alfvn wave damps . in the middle column of fig.5 parallel electric field @xmath124 is shown . at time @xmath125 , @xmath126 is zero everywhere , but as the alfvn wave propagates and its wave front becomes distorted @xmath126 starts to grow . it typically attains values of few @xmath127 . for solar coronal conditions this is typically exceeds the dreicer electric field ( which is associated with the particle acceleration runaway regime @xcite ) @xmath0 times ! hence , the generated parallel electric field efficiently accelerates electrons , as can be seen in the right column of fig.5 . the latter shows parallel to the magnetic field phase space ( @xmath128 vs. @xmath122 ) where each dot corresponds to an electron ( total of @xmath129 ) . we gather that in the regions where @xmath126 is generated , the number of particles with increased field - aligned velocity @xmath128 is increased ( i.e. particles are accelerated ) . note that this coincides with the regions where alfvn wave enhanced damping occurs . @xcite performed a parametric study , exploring how the new mechanism of parallel electric field generation and acceleration of electrons discovered by @xcite depends on problem parameters , such as the variation of frequency and amplitude of the applied ( driving ) alfvn wave , as well as plasma beta , affect levels attained by the @xmath126 and the fraction of accelerated particles ( the latter is defined as percentage of electrons with speeds above the electron thermal speed in the density gradient regions ) . it was established that ( i ) @xmath126 is always orders of magnitudes greater than dreicer electric field and ( ii ) the faction of accelerated electrons ranges within 20 - 50% . the latter makes the discovered electron acceleration mechanism quite efficient and potentially capable of resolving the above mentioned problems of standard flare models which use reconnection - produced electron beams instead of waves as the means of delivering energy flux to the footpoints in order to produce the x - rays . it should be noted that the transverse inhomogeneity is crucial in the model of @xcite ( see also further analysis paper @xcite ) . in fact , when the transverse density inhomogeneity is removed , i.e. when alfvn wave front remains always perpendicular to the uniform background magnetic field , no parallel electric field generation or electron acceleration is observed ( see figs.7 - 10 in @xcite ) . in terms of the above @xmath102 discussion , this can be explained as following : when the density across the magnetic field is inhomogeneous , despite the fact that initially @xmath98 as the alfvn wave front starts to deform @xmath102 is efficiently generated . when on the other hand density is constant everywhere then alfvn wave front is always at right angles to the magnetic field , i.e. alfvn wave propagates strictly along the field , and thus @xmath98 at all times . recently , @xcite revisited the problem of solar flare electron acceleration with alfvnic pulses . further , @xcite quantified the full details , namely , exploring how fraction of accelerated particles is affected by such crucial parameter as transverse length - scale @xmath130 of the alfvnic pulse . they found that significant fraction of electrons would be accelerated if @xmath131 is few meters or less . a clear distinction should be made between models of @xcite and @xcite . as explained above , in the case of former , @xmath102 is _ initially zero _ and it seems that such wave is more plausible to exist in nature . this is because it represents simple , ubiquitous alfvnic _ harmonic _ wave . recall that even in mhd approximation @xmath102 is also zero at all times . it is crucial in @xcite s model that plasma has density inhomogeneity mimicking transverse density structure of the coronal loop . it is this inhomogeneity that makes alfvn wave front to distort and create finite @xmath102 and hence generate @xmath126 . in the case of @xcite s model @xmath99 ( and hence @xmath103 ) _ from the start _ and there is no transverse density inhomogeneity . also , a pulse instead of harmonic have is considered . if we try to compare plausibility of the two models the following considerations come to mind . both models `` work '' i.e. generate large enough @xmath126 and accelerate sufficient proportion of electrons if the following are true : in @xcite s model a large scale alfvnic , harmonic wave initially has @xmath98 , interacts with the transverse density inhomogeneity which has a scale of @xmath132 . in @xcite s model ( see their eq.(28 ) ) alfvnic pulse with @xmath133 $ ] travels in the homogeneous plasma ( where @xmath134 is the perturbation amplitude ) . in effect , both models imply either small scale transverse inhomogeneity of the medium or small scale transverse wave number . it is unclear at this stage whether we have such small scale phenomena in solar corona . perhaps , in situ coronal probes or remote sensing from distances much small than sun - earth distance could shed some light , as this is currently possible in earth magnetospheric applications . however , until such times the jury on this topic will remain out . in summary , wave based models offer attractive solutions to the above mentioned outstanding problems in solar flare particle acceleration . it is yet to be clarified however , how small - scale ( kinetic - scale ) phenomena feeds into the large - scale ( mhd - scale ) phenomena . from the fundamental point of view , of course kinetic - scale modelling includes all the essential physics . however , realistically shortcomings of man - made even the largest parallel supercomputers will unlikely offer us the performance we need . e.g. a full 3d vlasov code @xmath135 grids in the 6-dimensional phase space ( 3 spatial dimensions and 3 velocity dimensions ) require 32 terabytes ( 32,000 gb ) of ram . given that spatial grid size is the debye length ( @xmath136 m in the solar corona ) , in 3d this offers rather uninterestingly small physical volume . thus , there is always a restriction imposed to consider a lower dimensional physical systems . despite significant progress in understanding of physical processes ongoing in solar atmosphere it is enough to take a look as nasa or esa webpages describing a number of previous and existing successful space missions , as well future mission plans there is still a long and exciting way to go ! as far as coronal heating problem is concerned , in author s opinion further progress could be made if following considerations are noted . from the observational perspective it would be rather important to investigate heating of active regions ( ars ) . ar is a volume of solar atmosphere above sunspots . they are important because during solar maximum ars can be responsible for @xmath137% of heating of the entire corona @xcite . top left of fig.6 shows trace 171 @xmath138 image of an ar . the box in fig.6 shows potential extrapolation ( using green s function method ) of magnetic field lines based on an input from the measured normal component of the photospheric magnetic field that can be used for numerical modelling . ideally , if we have _ simultaneous _ , high time cadence 2d imaging and 2d doppler shift data at several wavelength , @xmath139 , corresponding to emission from e.g. white light ( continuum , photosphere , @xmath140 ) , heii ( @xmath141 , chromosphere , transition region , @xmath142 ) , feix ( @xmath143 , quiet corona , upper transition region , @xmath144 ) fexii , fexxiv ( @xmath145 , corona and hot flare plasma , @xmath146 ) , etc then one would able to fourier transform the data to see e.g. how much wave power spectrum is lost ( dissipated ) from one height ( a horizontal cross - section through the box in fig.6 ) to another . essentially , different line forming ions probe different temperatures and therefore different heights in the box in fig.6 . this would ultimately enable us to make a judgement to what extent _ waves _ contribute to the ar and coronal heating ? also , measuring the abundance of the observed bi - polar jets , which are believed to be signatures of small - scale reconnection , could enable us to make a judgement on _ reconnection _ s contribution to the coronal heating . nasa s atmospheric imaging assembly ( aia ) on board of the solar dynamics observatory ( sdo ) , to be launched in november 2009 , will make such unprecedented progress . however , aia s resolution of about 1 arcsec and time cadence of 10 seconds or better will naturally limit the types of waves it can detect . perhaps a better job could be done by a ground based project called rapid oscillations in solar atmosphere ( rosa ) , @xmath147 , which will have 0.1 arcsec resolution and will take in 30 - 125 frames per second ! however , rosa being a ground based instrument , can only probe photosphere / chromosphere / tr lines , i.e. can not probe corona , and yet a whole wealth of new data is expected which will hopefully lead to new discoveries . on a negative note , with sdo s aia and rosa importance of kinetic effects still can not be probed , because 1 arcsec on the sun is @xmath148 m orders of magnitudes larger than any kinetic scale discussed above . the concept of measuring wave dissipation with height has been used recently by @xcite . this was based on variations in euv line widths of coronal plumes ( viewed off the solar limb ) with height . image of an active region in solar corona . the box shows potential extrapolation ( using green s function method ) of magnetic field lines based on an input from the measured normal component of the photospheric magnetic field that can be used for numerical modelling . ] also , there seems to be is a serious gap in the studies of the coronal heating problem , as here are no studies which consistently would monitor global _ ( average ) coronal temperature_. there are studies that monitor tsi ( total solar irradiance ) and its correlation to the ( 11 year- ) solar activity cycle . recent study by @xcite suggests that the long - term trend of tsi is most probably caused by a global temperature change of the sun ( photosphere ) that does not influence the uv irradiance in the same way as the surface magnetic fields . there is however , a work of @xcite , where they split the differential emission measure ( dem ) into 10@xmath149 sectors and estimate coronal heating requirement sector - by - sector . for every full - disk dataset with euv and sxr coverage , it would be straightforward to compute a dem of the entire corona , from which an emission measure - weighted temperature could be extracted as function of time . i.e. it could be possible to produce a graph of averaged ( by all sectors i.e. full disk ) coronal temperature as function of time , over the time interval of e.g. @xmath150 years with sufficient time cadence ( e.g. once per year or more ) . such plot would answer a significant question : whether the heat release in the corona is indeed related to the magnetic field ( which we know changes on 11 year timescale ) . there is certainly a good correlation between the magnetic flux and sxr intensity ( e.g. @xcite , or see fig . 1.13 in @xcite indeed , there seems to be a good case for @xmath151 ) . however , it is questionable whether @xmath152 is a good proxy for global average coronal temperature , because sxr emission is non thermal ( produced by the forbidden lines ) cf . fig.2.3 from @xcite showing strong deviation of coronal emission from the blackbody spectrum . the analogy with a patient who has caught cold is relevant here : if a patient is sick , its temperature is routinely monitored by doctors . if we claim there is a coronal heating problem , indeed , solar physicists should monitor the global average temperature of their patient the sun s corona ! as far as modelling challenges are concerned , it seems a real progress can be made if two - fluid simulation of heating release in ar is attempted . when ion and electron dynamics is decoupled , we know that reconnection will proceed fast @xcite even without invoking anomalous resistivity . what is encouraging is that we do not need to use realistic ion to electron mass ration of @xmath153 . as previous results have shown , as long as ion - electron dynamics is decoupled simulation with @xmath74 is as good @xcite . this implies that significant reduction in cpu requirements can be achieved . the two - fluid simulation results then can be compared with the sdo s aia and rosa data to see whether inclusion of the two - fluid effects can provide fast enough reconnection or generated wave power and hence adequate coronal heating . author is supported by the science and technology facilities council ( stfc ) of the united kingdom .	the solar corona , anomalously hot outer atmosphere of the sun , is traditionally described by magnetohydrodynamic , fluid - like approach . here we review some recent developments when , instead , a full kinetic description is used . it is shown that some of the main unsolved problems of solar physics , such as coronal heating and solar flare particle acceleration can be viewed in a new light when the small - scale , kinetic plasma description methods are used . [ firstpage ]
the detailed suppression pattern , @xmath6 , and elliptic flow , @xmath5 , of high - transverse - momentum hadrons is an important experimental signature of the quark - gluon plasma creation in heavy ion collisions @xcite . jet quenching for light mesons , such as @xmath8 , @xmath9 and @xmath10 , at rhic is well explained by radiative energy loss calculations @xcite . it also gives the dominant contribution to the azimuthal asymmetry of hard probes @xcite . in contrast , models @xcite with a physically reasonable set of qgp temperatures and densities , predict a qcd heavy - quark energy loss which is too small compared to the measured suppression of single non - photonic electrons @xcite . therefore , it is critical to investigate new interaction mechanisms in the qgp that may be specific to heavy flavor @xcite . thermalization of heavy quarks in the qgp - heat bath has been recently studied in the framework of the parton - transport approach @xcite . large nuclear suppression and elliptic flow @xmath11 result when employing a fokker - plank equation , @xmath12 \ ; , \label{fp}\ ] ] solved via an equivalent langevin simulation . in eq . ( [ fp ] ) @xmath13 is the distribution of @xmath3- and @xmath4-quarks and @xmath14 are the drag / diffusion coefficients , respectively . it has been argued that strong coupling between the @xmath3- and @xmath4-quarks and the qgp medium may be generated via quark - resonance interactions near the qcd - phase transition , @xmath15 @xcite . it is , therefore , important to study the interplay between such non - perturbative effects and the radiative @xcite and collisional @xcite heavy - quark energy loss . we evaluate @xcite the contribution of these processes to the drag and diffusion coefficients , @xmath16 which are then applied in the relativistic fokker - planck equation . for heavy @xmath3- and @xmath4-quarks @xcite from collisional @xcite and radiative energy loss @xcite and quark - resonance interactions @xcite . for charm quarks , the pqcd @xmath17 contribution is shown separately . right panel : elliptic flow @xmath5 for heavy c- and b - quarks for the same physics mechanisms @xcite . , title="fig:",width=234 ] for heavy @xmath3- and @xmath4-quarks @xcite from collisional @xcite and radiative energy loss @xcite and quark - resonance interactions @xcite . for charm quarks , the pqcd @xmath17 contribution is shown separately . right panel : elliptic flow @xmath5 for heavy c- and b - quarks for the same physics mechanisms @xcite . , title="fig:",width=234 ] results on the @xmath18-dependent suppression pattern of heavy quarks , @xmath6 , are shown in the left panel of fig . [ figure1 ] . drag coefficients are easily evaluated from the fractional momentum loss of heavy quarks , see eq . ( [ coefs ] ) . the diffusion coefficients in this preliminary study were constrained from the fluctuation - dissipation relation . we observe that the suppression of charm quarks can be very large even in minimum - bias reactions of large nuclei and @xmath19 . the high-@xmath18 azimuthal asymmetry for minimum - bias au+au collisions is shown in the right panel of fig . [ figure1 ] . we note that the generated @xmath11 for @xmath4-quarks is much smaller than that for @xmath3-quarks . one of the reasons for the large suppression in our current energy - loss implementation is that the einstein fluctuation - dissipation relation induces minimal gaussian fluctuations . these are significantly different from the ones in the probabilistic treatment of pqcd - energy loss @xcite . future langevin simulations of @xmath3- and @xmath4-quark diffusion should include momentum fluctuations beyond the einstein s relation and the decay of the heavy quark / hadron spectra into @xmath20 for direct comparison to the non - photonic electron observables at rhic @xcite . in the perturbative qcd - factorization approach , the cause of the limited single non - photonic electron quenching is identified as the small suppression of @xmath2-mesons , which dominate the high-@xmath18 @xmath21 yields . such models assume that the hard jet hadronizes in vacuum , having fully traversed the region of hot and dense nuclear matter , @xmath22 fm , and lost energy via radiative and collisional processes @xcite . in ref . @xcite we examined the validity of this assumption for different species of final - state partons and decay hadrons . for a @xmath23 gev pion at mid - rapidity @xmath24 fm @xmath25 , consistent with the jet - quenching assumptions @xcite . in contrast , @xmath2- and @xmath1-mesons of the same @xmath18 have formation times @xmath26 fm , respectively , @xmath27 . therefore , at the finite @xmath18 range accessible at rhic and lhc a conceptually different approach to the description of @xmath1- and @xmath2-meson quenching in a+a collisions is required , when compared to light hadrons . - and @xmath2-meson production via collisional dissociation in the qgp in central au+au and cu+cu reactions at rhic @xcite for gluon rapidity densities @xmath28 and @xmath29 , respectively @xcite . right panel : quenching of the non - photonic electrons from the softened @xmath1- and @xmath2-meson spectra in central au+au collisions @xcite . data is from phenix @xcite and star @xcite . , title="fig:",width=238 ] - and @xmath2-meson production via collisional dissociation in the qgp in central au+au and cu+cu reactions at rhic @xcite for gluon rapidity densities @xmath28 and @xmath29 , respectively @xcite . right panel : quenching of the non - photonic electrons from the softened @xmath1- and @xmath2-meson spectra in central au+au collisions @xcite . data is from phenix @xcite and star @xcite . , title="fig:",width=238 ] motivated by this finding , in the framework of the glv theory , we derive the collisional dissociation probability of heavy mesons in the qgp @xcite : @xmath30 \geq 0 \ ; , \quad p_d ( \chi\mu^2 \xi = 0 ) = 0 \;. \label{sprob}\end{aligned}\ ] ] in eq . ( [ sprob ] ) @xmath31 is the cumulative 2d transverse momentum squared per parton . the dissociation probability also depends on the detailed heavy meson light cone wave function . the dynamics of open heavy flavor production and modification in this model is represented by a set of coupled rate equations that describe the competition between @xmath4- and @xmath3-quark fragmentation and @xmath1- and @xmath2-meson dissociation @xcite : @xmath32 in eqs . ( [ rateq1 ] ) and ( [ rateq2 ] ) @xmath33 . for further details , see @xcite . we solve this system of coupled rate equations numerically , using the same initial soft - gluon rapidity density @xmath34 as in the calculation of the @xmath35 quenching @xcite in central au+au and cu+cu collisions at rhic . our results , including a study of a range of anticipated qgp densities at the lhc , are shown in the left panel of fig . [ figure2 ] . contrary to calculations that emphasize radiative and collisional heavy quark energy loss @xcite , qgp - induced dissociation predicts @xmath2-meson suppression comparable to or larger than that of @xmath1-mesons at transverse momenta as low as @xmath7 gev @xcite . the heavy meson spectra are decayed into electrons @xmath36 using the pythia event generator . our results are shown in the right panel of fig . [ figure2 ] . the predicted @xmath6 , which does not neglect the large @xmath2-meson contribution , describes well the most recent heavy flavor quenching measurements at rhic @xcite . we emphasize that such agreement between theory and experiment is not achieved at the cost of neglecting the contribution of the @xmath2-mesons to the non - photonic @xmath21 spectra . in these proceedings , we compared two new theoretical approaches @xcite to open heavy flavor modification in the qgp . preliminary results on langevin simulations of heavy quark diffusion , which include radiative energy loss , collisional energy loss and quark - resonance interactions , were shown . while an improved implementation of momentum fluctuations is required for quantitative comparison between data @xcite and theory @xcite , we find normal suppression , @xmath37 , and elliptic flow , @xmath38 , hierarchies as function of the heavy quark mass . results on qgp - induced collisional dissociation of heavy mesons @xcite were also shown . a good description @xcite of the large quenching of the inclusive non - photonic electrons @xcite is achieved by this model . a natural consequence of the approach developed in ref . @xcite is that @xmath2-mesons are attenuated as much as @xmath1-mesons at transverse momenta as low as @xmath7 gev . we conclude that robust experimental determination of the dominant mechanism for in - medium modification of open heavy flavor would require direct and separate measurements of the @xmath2- and @xmath1-meson @xmath39 and @xmath11 distributions versus @xmath18 and centrality in collisions of heavy nuclei . * acknowledgment : * this work is supported by the u.s . department of energy under contract no . de - ac52 - 06na25396 ( iv ) and grant no . de - fg02 - 93er40764 ( aa ) and by the u.s . national science foundation under grant no . phy-0449489 ( hvh ) .	we revisit the question of the measured , unexpectedly large , heavy flavor suppression , @xmath0 , in nucleus - nucleus collisions at rhic and compare two new theoretical approaches to the @xmath1- and @xmath2-meson quenching . in the first model , radiative energy loss , collisional energy loss and heavy quark - resonance interactions are combined to evaluate the drag and diffusion coefficients in the quark - gluon plasma and the mixed phase . these are applied in a relativistic fokker - planck equation to simulate the heavy @xmath3- and @xmath4-quark suppression rate and elliptic flow @xmath5 . in the second model , the fragmentation probability for heavy quarks and the medium - induced decay probability for heavy hadrons are derived . these are implemented in a set of coupled rate equations that describe the attenuation of the observable spectra from the collisional dissociation of heavy mesons in the qgp . an improved description of the non - photonic electron @xmath6 at rhic can be obtained . in contrast to previous results , the latter approach predicts suppression of @xmath2-mesons comparable to that of @xmath1-mesons at transverse momenta as low as @xmath7 gev .
the relative abundances of the light elements not only substantiate the standard interpretation for big bang nucleosynthesis ( bbn ) ( reeves et al . 1973 ; epstein , lattimer , & schramm 1976 ) , but they also hold the key for our determining the universal ratio of baryons to photons , commonly designated by the parameter @xmath10 ( boesgaard & steigman 1985 ; olive et al . 1990 ; smith , kawano , & malaney 1993 ) . there has been considerable interest in measuring the abundance of deuterium , since its production was strongly regulated by photodestruction in the radiation bath during the bbn , making d / h a strong discriminant of @xmath10 . deuterium is also destroyed in stars . after having passed through one or more generations of stars , diffuse gases that we can observe have probably had their deuterium abundances reduced to values below those that result from bbn . thus it is important to observe systems that have different levels of chemical enrichment and mixing ( timmes et al . 1997 ) , so that we can untangle the effects of the two fundamental destruction mechanisms , i.e. , the photodestruction accompanying bbn and the astration of material as the universe matures . a key step in this area of research is to form a solid foundation of measurements of d in the chemically evolved gas in the disk of our galaxy . ultimately , when these results are combined with determinations for distant gas systems that have not aged as much , we expect to achieve a better understanding about the processing of gas through stars , which is interesting in its own right ( steigman & tosi 1992 , 1995 ; dearborn , steigman , & tosi 1996 ; scully et al . 1997 ; tosi et al . 1998 ) , and this in turn should allow us to extrapolate the concentration of d back to an era very soon after its primordial production . an important foundation in recognizing the relationship between d / h and some measure of stellar processing , such as the relative abundances of elements produced in stellar interiors , is that an empirical relationship between the two forms a unique sequence . if this turns out not to be true , then more elaborate interpretations of chemical evolution may be needed . to explore this issue , we have embarked on a program to revisit some lines of sight studied by other investigators ( [ bkgnd ] ) , but this time using much higher resolution spectra obtained with the interstellar medium absorption profile spectrograph ( imaps ) . in this paper , we investigate the spectrum of @xmath1 ori a ( hd 36486 ) . this star has a spectral classification of o9.5 ii , is a spectroscopic binary , and is a member of the ori ob1 association that has a distance modulus of 8.5 ( @xmath11 ) ( humphreys 1978 ) . in a companion paper ( sonneborn et al . 1999 ) we will report on results for @xmath12 vel and @xmath13 pup . the basic properties of our spectrum of @xmath1 ori and the instrument that recorded it are discussed in [ basic ] , followed by examinations of systematic errors that could arise in the determination of a very weak contamination signal ( [ contam ] ) , the intensity of the scattered light background ( [ bkg ] ) , and absorption features from other species ( [ interference ] ) . in [ template ] we describe how we obtained independent information on the shape of the velocity profile for material toward @xmath1 ori , so that we could undertake our study with only a small number of unknown , free parameters . we have paid considerable attention to minimizing the errors and evaluating them in a fair and consistent manner ( [ errors ] ) . for our value of @xmath8(d i ) reported in [ n(di ) ] to be useful , we must compare it with @xmath8(h i ) , and we must strive to make the accuracy of the latter as good as or better than the former . in [ n(hi ) ] we discuss our comprehensive investigation of the iue archival data that show l@xmath5 absorption in the spectrum of @xmath1 ori . this special analysis combined with our determination of @xmath8(d i ) ultimately led to our determination of the atomic d / h toward @xmath1 ori reported in [ d / h ] . we relate this result to the abundances of other elements relative to h in [ heavy ] and discuss its significance in [ disc ] . early measurements of d / h obtained from the _ copernicus _ satellite ( resolution @xmath14 fwhm ) and iue ( resolution @xmath15 fwhm ) , summarized by vidal - madjar & gry ( 1984 ) , had a few cases that differed by more than the reported errors from a general average @xmath16 . at face value , this suggested that d / h varies from one location to the next . mccullough ( 1992 ) revisited this problem and asserted that the evidence for such variations was not convincing . in making his claim that all of the data were consistent with a constant value for d / h , mccullough rejected all of the deviant cases on the grounds that the complexity of their velocity structures made the measurements much less accurate than originally claimed . the high resolution ( @xmath17 fwhm ) and good sensitivity of the ghrs instrument on hst enabled an accumulation of very accurate observations of the interstellar l@xmath5 h and d absorption features superposed on the broader chromospheric l@xmath5 emission lines of nearby f , g and k type stars . the best determinations of d / h were those toward @xmath5 aur ( capella ) , where linsky , et al . ( 1995 ) found that @xmath18 , and toward hr 1099 , where piskunov , et al . ( 1997 ) obtained @xmath19 . the issue of whether or not atomic deuterium to hydrogen ratios toward other cool stars differ from these values has been an elusive one , although it seems clear that one could rule out deviations greater than about 50% in either direction ( wood , alexander , & linsky 1996 ; dring et al . 1997 ; piskunov et al . the chief problem has been that the measurements of @xmath20 toward late - type stars were very dependent on assumptions about the shape of the underlying emission profile ( linsky & wood 1996 ; piskunov et al . 1997 ) or the compensations for additional , broad absorptions caused by hydrogen walls associated with the stellar wind cavities around either the sun or the target stars ( linsky & wood 1996 ; wood & linsky 1998 ) . even so , these investigations revealed some intriguing , convincing variations for the abundances of d i with respect to those of mg ii . unfortunately , the significance of these changes is clouded by the possibility that they could result simply from alterations in the amount of depletion of mg onto dust grains ( murray et al . 1984 ; jenkins , savage , & spitzer 1986 ; sofia , cardelli , & savage 1994 ; fitzpatrick 1997 ) . lemoine , et al ( 1996 ) observed the interstellar h and d l@xmath5 absorption features in the spectrum of the da white dwarf g191@xmath21b2b with the ghrs and reported their determinations for d / h . later , high - resolution observations by vidal - madjar , et al ( 1998 ) brought forth some refinements in the interpretation of the velocity structures of the absorption profiles , leading to a determination @xmath22 for all of the material in front of this star . if one allows for the fact that a contribution from the local interstellar cloud ( lic ) is somewhat blended with those of more distant clouds and adopts the @xmath5 aur result for the lic , d / h toward the other material could be of order @xmath23 . this low value for d / h is supported by observations of the hot subdwarf bd @xmath24 reported by glz , et al . ( 1998 ) , @xmath25 , although the error bar is large enough to include the results obtained for @xmath5 aur , hr 1099 , and other late - type stars . for lines of sight that have hydrogen column densities that are small enough to analyze using the l@xmath5 profile , there is the danger that improper allowances for either l@xmath5 emission ( cool stars ) or absorption ( hot dwarfs ) could lead to errors . moreover , in some circumstances hydrogen walls associated with either the target stars or the sun can lead to complications . one way to bypass these problems is to examine the higher lyman series absorption features toward more distant , early - type stars with much more foreground material , as was done with the _ copernicus _ satellite . we also have the benefit of sampling the interstellar medium well outside our immediate vicinity . however , a principal weakness of _ copernicus _ was its limited resolving power ( @xmath26 fwhm ) . per , an effect which they attributed to the influence of stellar features that varied with time . ] in large part , the _ copernicus _ investigators had to model the instrumentally smeared , detailed velocity structure of the gas , with guidance from high - resolution observations of na i features recorded from the ground . unfortunately , the sodium d lines are a poor standard for comparison because their strengths are dependent on ionization equilibria that are entirely different from those of d i and h i. in this study we revisit the case for @xmath1 ori , originally observed with _ copernicus _ by laurent , et al . ( 1979 ) , but now with new observations taken with an instrument with considerably better velocity resolution than a far - uv spectrum of @xmath1 ori over the wavelength interval 930 to 1150 was recorded in a series of exposures lasting 54 min over various observing intervals between 22 november and 3 december 1996 by the interstellar medium absorption profile spectrograph ( imaps ) . this series of observations was undertaken during the orfeus - spas ii mission ( hurwitz et al . 1998 ) on sts-80 , which was the second orbital flight of imaps . imaps is a simple , objective - grating echelle spectrograph that can record the far - uv spectrum of a bright , early - type star with sufficient resolution to show many of the velocity structures in the interstellar lines . jenkins , et al . ( 1996 ) present a detailed description of the imaps instrument , its performance on the first orfeus - spas mission in 1993 , and the methods of data correction and analysis . we summarize very briefly how the spectra are recorded by imaps : in any single exposure that covers an angle @xmath27 , one - quarter of the echelle grating s free spectral range and , nominally , diffraction orders 194 through 242 are recorded by an electron - bombarded ccd image sensor . this detector has an opaque photocathode on a smooth substrate and uses magnetic focusing to form the electron image on the ccd . electrons impacting on the back side of the specially thinned ccd have an energy of 18 kev , and they produce enough secondary electrons within the silicon layer to make individual photoevents appear as bright spots . each spot has an amplitude that is about 20 times greater than the combined noise from the readout amplifier and the random fluctuations in dark current . the ccd has a format of @xmath28 pixels , each of which is @xmath29 m square and subtends a @xmath30 equivalent to a doppler shift of @xmath31 . the echelle orders are separated by about 5 ccd pixels , but they are rather broad in the cross - dispersion direction . the ccd is read out 15 times a second . the video signals from successive frames are summed in an accumulating memory to produce the integrated spectral images . in our processing of these images after the flight , we subtracted dark - current comparison frames that were recorded at frequent intervals with the accelerating high voltage turned off . the effective area of imaps on the 1996 flight was about @xmath32 at wavelengths longward of about 1020 , leading to typical signal - to - noise ratios of about 80 near the maximum of the echelle grating s blaze angle for stars as bright as @xmath1 ori a. however the al+lif coatings on the two gratings have a low reflection efficiency at shorter wavelengths , resulting in a factor of 10 lower effective area in the vicinity of the l@xmath1 and l@xmath2 lines . this reduced efficiency coupled with the much lower flux at the centers of the stellar lyman series lines made it especially difficult to achieve high values of signal - to - noise . we overcame this problem by recording a large number of spectra that could be added together . the total integrated flux at the continuum levels near l@xmath1 and l@xmath2 amounted to about 600 photons for each ccd pixel width in the dispersion direction ( @xmath33 ) . noise fluctuations in the spectra had @xmath34 deviations about equal to 1/10 of the continuum levels , with the principal noise source being the multiple readouts of the ccd , rather than from photon - counting statistical errors . ( this is clearly evident in fig . [ dorideut ] , which shows a noise level at zero intensity to be about the same as that at the elevated intensity levels . ) we deliberately introduced offsets in position for the spectra in different sets of exposures . this was done to reduce the possibility that the spectrum could be perturbed by subtle flaws , such as ccd columns with anomalous responses or variations in photocathode efficiency with position ( we could see no evidence for the latter however ) . figure [ dorideut ] shows the d and h absorption profiles for @xmath1 ori at l@xmath1 and l@xmath2 . observations of telluric atomic oxygen lines from excited fine - structure levels , seen elsewhere in the imaps spectrum of this star , indicated that the instrumental profile that governs the wavelength resolution of these observations was consistent with a gaussian distribution having a fwhm ori a are not far from those that apply to our @xmath1 ori spectrum . ] equal to @xmath35 . at this resolving power , the deuterium features are well separated from their hydrogen counterparts , as can be seen in fig . [ dorideut ] . while the deuterium l@xmath1 profile does show some asymmetry , to within the uncertainties of the noise there does not seem to be any extraordinary complexity in the velocity structure of the d profiles . for instance , there seems to be no evidence for any strong , narrow velocity components buried within the main peak of the l@xmath2 feature . most important , the d features are not badly saturated . there is some overlap of signal from one echelle diffraction order to the next . our optimal extraction routine was designed to compensate for this effect ( jenkins et al . 1996 ) , but if this correction was not perfect we may have had some contamination by the spectral intensities from an adjacent order . we had to be especially watchful for this possibility in the vicinity of the lyman series lines because the stellar continuum level in the region of interest is much lower than elsewhere . this gives the contamination signal an advantage over the signal we wished to study . for l@xmath2 there is no problem because there are no spectral features in the orders either directly above or below the d and h features or their nearby continua . errors in correcting for order overlap will only change the effective zero intensity level , which is corrected out anyway . the next higher order of diffraction that appears just below the one that contains the l@xmath1 absorptions is featureless , but , unfortunately , the order above the l@xmath1 order exhibits interstellar absorption features from the very strong multiplet of n i at 954 . in our investigation of the spectrum in the vicinity of l@xmath1 , we allowed for the possibility that our correction for the order overlap was either too large or too small . this error could have added a spurious signal to our spectrum . therefore , we included as a free parameter a scaling coefficient @xmath36 ( which could be either positive or negative ) for the amplitude of a correction signal ( with the same shape ) to cancel the possible residual contamination , and we allowed it to vary as we explored for minimum values of @xmath3 ( [ errors ] and [ n(di ) ] ) . when this coefficient is less than 0.005 or larger than 0.03 , unreasonably large perturbations can be seen in the bottom of the very broad hydrogen feature . within this range , however , we allowed for the fact that our derivation of @xmath8(d i ) could be influenced by the exact value . figure [ dorideut ] shows the correction signal with the most plausible amplitude , as indicated by minimum @xmath3 for @xmath36 at the most probable @xmath8(d i ) given in table [ chi2_results ] . the spectrum that is shown in this figure has had this correction included . our analysis of the d profiles is very dependent on our having an accurate determination of the level of zero intensity . sources of background illumination include not only grating scatter , but also a diffuse glow caused by a portion of the l@xmath5 geocoronal background that is not fully rejected by a mechanical collimator at the instrument s entrance aperture . ( the detector s dark count rate is negligible compared to these sources of background . ) fortunately , we could use the bottoms of the broad , heavily saturated h absorptions that accompany each d profile establish the position of the background level . in principle , the saturated portion of the h profile could mislead us if there were broad , shallow wings in the instrumental profile caused by scattering from the echelle grating . if this were the case , one could imagine that the local background level might increase slightly for wavelengths somewhat removed from the strong h feature . we can rule out this prospect on two grounds . first , before imaps was flown , we illuminated it in a vacuum tank with a collimated beam from a molecular hydrogen emission line source , and faint , very broad wings of the recorded emission lines could be seen only on the strongest features . the energy in these wings corresponded to 15% of the total , spread over several . the remaining 85% was within the main peak . second , for @xmath1 ori we found that for both l@xmath1 and l@xmath2 the apparent depths of the d features in _ copernicus _ scans ( taken with an ordinary grating in first order ) showed excellent agreement with those registered in the imaps spectrum after it had been degraded by convolving it with the _ copernicus _ instrumental profile function [ a triangle with fwhm = 0.045 ( laurent , vidal - madjar , & york 1979 ) ] . for these two reasons , we feel confident that the apparent flux in the bottom of the h feature is , to within the uncertainties from noise , a good representation for the zero level under the deuterium line . table [ other_lines ] lists lines from species other than d i and h i that are in vicinity of the deuterium absorptions or the fragments of the spectrum that were used to define the continuum level ( [ n(di ) ] ) . the werner 3@xmath210 p(2 ) line lies within the h l@xmath1 feature , and thus it is of no importance . from other lines out of the @xmath37=4 level of h@xmath38 that appear elsewhere in our imaps spectrum , we know that the 4@xmath210 p(4 ) line of the werner system should have a negligible strength . again using other features in the imaps spectrum , we found that the remaining two lines shown in the table could perturb our spectra and influence our final results . we therefore felt it was necessary to estimate their strengths and then apply a correction to compensate for their presence . lccl h@xmath38 werner 3@xmath210 p(2)&949.608&@xmath39&within the l@xmath1 h i absorptionh@xmath38 lyman 14@xmath210 p(2)&949.351&0.95&on cont . left of the l@xmath1 d i absorptionfe ii&937.652&1.263&on cont . between l@xmath2 d i and h i absorptionsh@xmath38 werner 4@xmath210 p(4)&937.551&0.89&too weak to matter to estimate the strength of the 14@xmath210 p(2 ) lyman line of h@xmath38 , we noted that the 4@xmath210 r(2 ) line at 1051.498 had a maximum depth of 0.27 times the local continuum at @xmath40 , and it was recorded in a part of our spectrum where the signal - to - noise ratio was about 80 . this line has a value for @xmath41 that is 1.7 times that of the 14@xmath210 p(2 ) line . to compensate for the effect of the latter on our continuum to the left of the d i l@xmath1 feature , we divided the observed spectrum by the continuum - normalized intensities in lyman 4@xmath210 r(2 ) profile all taken to the 1/1.7 power , with the profile shifted in wavelength to match that of the 14@xmath210 p(2 ) line . for a template of the fe ii absorption , we used the line at 1081.875 that shows a maximum depth of 0.26 at @xmath42 recorded at s / n = 90 ( this maximum for the 937.652line falls within the h absorption ) and a shoulder at @xmath43 with a depth of 0.12 . this shoulder for the 937.652 line falls on top of a critical piece of continuum between the d i and h i l@xmath2 features . the 1081.875 line has @xmath41 that is 0.82 times that of the interfering feature ( morton 1991 ) , and once again this difference was taken into account when we made the correction . to derive the most accurate value for the column density of atomic deuterium @xmath8(d i ) , it is beneficial to use information from other species recorded at much higher s / n to help define more accurately the shape of the d i velocity profile . profiles of o i and n i are probably the most suitable comparison examples for two reasons . first , these two elements have very mild depletions , if any , caused by the atoms condensing into solid form onto dust grains ( meyer , cardelli , & sofia 1997 ; meyer , jura , & cardelli 1998 ) . the column densities of n and o seem to track those of h over a diverse sample of regions ( ferlet 1981 ; york et al . 1983 ) . as a consequence , it is unlikely that their velocity profiles will differ appreciably from that of d i. this is in contrast to the usual striking differences exhibited between elements that are mildly depleted , such as na i , and elements that are strongly depleted , such as ca ii . the former is generally concentrated at lower velocities than the latter for a given line of sight ( routly & spitzer 1952 ; siluk & silk 1974 ; vallerga et al . 1993 ; sembach & danks 1994 ) . second , o i and n i have ionization potentials close to that of neutral hydrogen , and this close match in energy makes their susceptibility to ionization nearly the same and also insures that the cross sections for ( nearly resonant ) charge exchange are high ( field & steigman 1971 ; butler & dalgarno 1979 ) . for this reason , plus the consideration that whatever means there are for ionizing n and o will operate in much the same way for h ( or d ) , we can generally regard the relative ionizations of oxygen and nitrogen to be good representations for that of d [ but for evidence to the contrary , see vidal - madjar et al . ( 1998 ) ] . this assumes , of course , that o and n are not being ionized appreciably to multiply charged states . ori a. no absorption was evident at @xmath44 , but it was difficult to assign a quantitative upper limit because of interference from the nearby feature of si ii at 989.873 . ] in the wavelength coverage of imaps where we have a reasonably good s / n , there are no transitions from the o i ground state that are weak enough ( and with known f - values ) to yield absorption lines that we can analyze . there is , however , a good series of exposures in the hst archive that cover the o i 1355.6 feature in the spectrum of @xmath1 ori , recorded by the ghrs echelle - a spectrograph . unfortunately , the transition probability for this line is so weak that only the main peak in the velocity profile shows up above the noise . for n i , within the coverage of imaps there are three multiplets ( at 952.4 , 953.8 and 954.1 ) from the ground level that are ideal for studying the apparent distribution ori at high velocity resolution , it is probably safe to assume that the two are equal to each other . ] @xmath45 of the nitrogen atoms with velocity , defined as @xmath46 where the apparent optical depth @xmath47 is a valid quantity to measure at velocities @xmath48 where the line is not badly saturated or , alternatively , not too weak . in this equation @xmath49 is the transition s oscillator strength , and @xmath50 is expressed in . in our study of the n i lines , we adopted f - values from the laboratory measurements of goldbach , et al . for the triplet at 952.4 , the weakest feature at 952.523 is only moderately saturated . the other two features are heavily saturated but useful for revealing the weaker shoulder on the left - hand side of the main peak . the 4 much stronger features of n i in the vicinity of 954 are useful for defining accurately the behavior of @xmath45 at velocities where it is below about @xmath51 . we derived a composite @xmath45 profile for n i from the 7 lines using the method employed by jenkins & peimbert ( 1997 ) when they synthesized the profiles of h@xmath38 in various @xmath37 levels toward @xmath13 ori a. figure [ nando ] shows the @xmath52 profiles that we derived for n i and o i. we chose to work with the n i profile in our interpretation of the d lines because it was of much better quality . this pragmatic reason for choosing n i as a template is contrary to the idealistic stance that o i would be a better match to d i , based on evidence from the absorption lines in the spectrum of g191-b2b recorded by vidal - madjar , et al ( 1998 ) and calculations of partially ionized atomic gases by sofia & jenkins ( 1998 ) . we note , however , that to within the noise fluctuations the significant part of the profile of o i is consistent with the main part of the profile of n i. there is a suggestion of an apparent inconsistency between n i and o i in the velocity range @xmath53 . while this may be true , we point out that this weak shoulder in the o i absorption , if it exists , may have been lost in the fitting of the continuum to the curvature of the stellar spectrum , which is larger than the expected size of the shoulder indicated by the n i profile . also , the existence of some absorption to the left of a main peak is supported by the shape of the l@xmath1 deuterium profile shown in figures [ dorideut ] and [ extremes ] . in principle , we should be cautious about possible contamination of the interstellar n i profile by nitrogen atoms in the earth s atmosphere above our orbital altitude of 295 km . above this altitude , the msis-86 model atmosphere for solar minimum shows an exponential decrease in the density of nitrogen atoms with a scale height of 53 km , starting at a density of @xmath54 at 295 km ( meier 1991 ) . for a zenith angle @xmath55 of 45 , we calculate that the telluric contribution to the observed @xmath8(n i ) should be @xmath56 , an amount that would be just about invisible in the representation shown in fig . [ nando ] . even at @xmath57 , @xmath58 , which is just slightly larger than the bump ( presumably due to noise ) immediately to the right of the `` ( n i)@xmath59 '' indication in the figure . thus , we feel it is safe to dismiss the possibility that any telluric n i contamination is large enough to influence our profile . one important difference between d and either n or o is the atomic mass . if the thermal doppler broadening for deuterium atoms is not very much less than that due to macroscopic motions , we would expect the d profiles to be broader than those of o and n. with the simplifying assumption that the temperature of the gas does not vary much from place to place , we expect that for nitrogen the observed distribution of the atoms with velocity is represented by the turbulent motions @xmath60 convolved with the thermal doppler profile , @xmath61 , given by @xmath62~,\ ] ] with @xmath63 equal to 14 times the proton mass @xmath64 . for convenience , we can include in @xmath60 the instrumental smearing of the profile , on the condition that we are not being misled by saturated , unresolved structures in the absorption line ( savage & sembach 1991 ; jenkins 1996 ) . the same relation holds for deuterium with @xmath65 . since the convolution of two gaussian distributions produces a third with a second moment equal to the sum of the two original ones , we can state that @xmath66 when we analyzed the deuterium features , we adopted for a standard model of their shapes the nitrogen velocity profile convolved with the last term in eq . [ convol ] . we allowed the temperature @xmath67 to be a free parameter that could influence the fit between the profiles of n i and d i ( and one that is also of some astrophysical relevance ) . we did not allow for variations of @xmath67 among unrecognized and blended velocity substructures that contributed to the profile , since the identification of these components is somewhat arbitrary . our goal was to account for the general modification of the profile due to the known differences in the effects of thermal and turbulent broadening . ( for determining @xmath8(d i ) , the weakness of the dependence of the derived @xmath8(d i ) with @xmath67 expressed in the endnote of table [ chi2_results ] indicates that our simplification that @xmath67 is constant is probably safe . ) figure [ nando ] shows our model for the deuterium profile ( smooth , dashed line ) for a value of @xmath67 that gave the minimum @xmath3 for the preferred value of @xmath8(d i ) in table [ chi2_results ] . this is a smoothed version of the n i profile ( heavy , solid line ) that was obtained from the convolution by the kernel @xmath68 from eq . [ convol ] . in addition to allowing @xmath67 to vary , we also allowed for the existence of a uniform velocity offset between n i and d i , in recognition of the possibility that either our wavelength scale or the laboratory wavelengths of the n i features had some small , systematic errors . the presence of random errors due to noise fluctuations in the signal presents the usual challenge of determining the most probable result for @xmath8(d i ) and permissible variations that still give an acceptable fit to the data . on top of this we must consider additional uncertainties caused by systematic errors . the ones that we can identify easily are the inaccuracies in defining the background level for both lines ( [ bkg ] ) and the contamination signal in the l@xmath1 profile([contam ] ) . additional parameters that could affect the outcome are the temperature of the gas @xmath67 through its effect in making the deuterium profile smoother than the n i template ( [ template ] ) , the difference in the zero points of the n i and d i velocity scales , and the adopted heights and slopes of the continuum levels over the deuterium l@xmath1 and l@xmath2 absorption features . our tactic for coping with these systematic errors was to express them in terms of simple parameters that could vary and then consider them in a unified analysis . since these errors could be correlated , we felt that it would be unwise to analyze them separately . we determined how well the data conform to various combinations of parameter values by evaluating the conventional @xmath3 statistic , @xmath69 ^ 2 $ ] , where @xmath70 is a measured intensity with uncertainty @xmath71 , and @xmath72 is the expected intensity given the specified set of parameter values . useful discussions of how to interpret the determinations of @xmath3 when there are many free parameters are given by lampton , et al . ( 1976 ) and bevington & robinson ( 1992 , p. 212 ) . the basic scheme is to find the minimum value @xmath73 that arises when all parameters that influence @xmath72 are free to vary , and then examine the deviations @xmath74 as the parameters stray from their optimum values . our @xmath3 values represented a sum over both the l@xmath1 and l@xmath2 features taken together . this approach is similar to one adopted by burles & tytler ( 1998a ) in their analysis of the deuterium abundance in a quasar absorption line system , except that we decouple the hydrogen measurements ( [ n(hi ) ] ) from those of deuterium because they are fundamentally different from each other . to limit the number of degrees of freedom that apply to the confidence intervals for the outcomes that we are interested in , we segregated the free parameters into two fundamental categories . first , we recognized those parameters that had an astrophysical significance , @xmath8(d i ) and @xmath67 . we sought to find a confidence interval that constrained these two parameters simultaneously . while @xmath67 might seem to be an incidental parameter outside the objective of this study , there were good physical reasons for our verifying that no appreciable portion of the probability density wandered above or below acceptable limits . the second category contained parameters that were of no particular interest to us , i.e. , nuisance parameters , but ones that had to be allowed to change freely as we re - minimized the @xmath3 for every new trial combination of @xmath8(d i ) and @xmath67 . we had no profound reason to require that any of these variables in the second category be constrained , and thus we could consider a projection of the lowest @xmath3 values in the multi - dimensional space of these variables onto just the @xmath8(d i)@xmath75 plane . this allowed us to restrict the number of the degrees of freedom ( df ) that applied to the confidence intervals down to only 2 . in summary , parameters that mattered were ( 1 ) @xmath8(d i ) and ( 2 ) @xmath67 , while those that did not were ( 3 ) the coefficient @xmath36 for scaling the contamination signal ( [ contam ] ) , ( 4 ) a relative velocity error @xmath76 between the n i profile template ( [ template ] ) and the deuterium absorption , ( 5 and 6 ) the background levels in the bottom of the h l@xmath1 and l@xmath2 features , and finally ( 7 through 10 ) the two coefficients that described the continuum straight lines spanning the d l@xmath1 and l@xmath2 features , i.e. , in each case the level near the middle of the d feature and the slope of the line.(h i ) given in [ n(hi ) ] , the damping wing absorbs only 9% of the flux at the d l@xmath1 line and 2% at the d l@xmath2 line . moreover , the _ appearance _ of the continuum suggests that a straight - line fit is justified see fig . [ extremes ] . ] from our knowledge of the ccd readout noise and dark current combined with the statistical fluctuations in the ( background + signal ) photons , we made an initial estimate for the uncertainties in the individual measurements of intensity @xmath71 at each velocity . we determined the correlation length for these errors by comparing fluctuations of intensity at a velocity @xmath48 with those of @xmath77 . the correlations disappear for @xmath78 , which is exactly the width of each pixel in the ccd .. ] our determinations of @xmath3 discussed below relied on intensities separated by this value for @xmath79 . the most important terms in the summation for @xmath3 are those that are directly influenced by trial values of @xmath8(d i ) and @xmath67 , through differences over the ( deuterium line ) velocity interval @xmath80 to @xmath81 between measured intensities @xmath70 and the computed values of @xmath72 , the expected absorption profile multiplied by a local continuum level . at the same time , parameters that define the continuum contribute to @xmath3 through the deviations between @xmath70 and straight - line extrapolations over the intervals ( @xmath82 to @xmath80 , + 40 to @xmath83 ) for l@xmath1 and ( @xmath84 to @xmath80 , + 40 to @xmath85 ) for l@xmath2 ( see fig [ extremes ] ) . likewise , @xmath3 is influenced by modifications in the background zero level that must be subtracted from the raw intensities at all velocities : we allowed the sum to include deviations away from zero for the background - corrected fluxes over the heavily saturated portion of the h profile from + 60 to @xmath86 ( for the d line heliocentric velocity scales shown in figs . [ dorideut ] and [ extremes ] ) . finally , fluxes over all velocities in the l@xmath1 profile can be modified by the contamination correction signal , whose amplitude was allowed to vary as we searched for a minimum @xmath3 in each case . we had a total of 320 independent intensity measurements to constrain the 10 free parameters listed at the end of [ errors ] , so we should insist that the minimum @xmath3 agree with a reasonable expectation for @xmath87 . in fact , with our original estimate for the noise in the measurements , we arrived at a minimum @xmath88 , a value that was unreasonably low . in later calculations , we rescaled this noise level by a factor @xmath89 , since we had a 90% confidence that the minimum @xmath3 should be at least equal to 278 for df = 310 . we felt that it was legitimate for us to perform a _ post facto _ rescaling of the noise , because our original estimate was accurate to only a level of about 25% . this rescaling is a conservative one , because it s more probable that the minimum @xmath3 should really be about equal to 310 . if we had used 310 instead of 278 in the expression for the noise multiplication factor , we accordingly would have found tighter limits for @xmath8(d i ) because the @xmath3 expressions would have increased more rapidly as we deviated away from the most probable @xmath8(d i ) . to find the minimum @xmath3 , we used powell s method of converging to the minimum of a multi - dimensional function ( press et al . 1992 , p. 406 ) . after finding this minimum and noting the most probable @xmath8(d i ) , we then evaluated the confidence interval for @xmath8(d i ) by forcing this parameter to vary , but at the same time allowing the other 9 parameters to adjust to new minima in @xmath3 . our target values for the new minima corresponded to @xmath90 and @xmath91 for the 90% and 99% confidence limits ( i.e. , `` 1.65@xmath92 '' and `` 2.58@xmath92 '' deviations ) , respectively , where @xmath93 is the overall minimum at the preferred value of @xmath8(d i ) as shown in table [ chi2_results ] . this exercise ultimately led to the limiting values for @xmath8(d i ) listed in the table . over the full range of @xmath8(d i ) between the most extreme limits , the temperature @xmath67 was the only parameter that showed any profound change . for this reason , @xmath67 is also listed . our result for the most probable @xmath8(d i ) is in near perfect agreement with the value @xmath94 reported by laurent , et al ( 1979 ) in their investigation that led to a value d / h = @xmath95 using data from _ copernicus_. lccc minimum @xmath8(d i ) at the 99% confidence limit&0.89&6800&286.1minimum @xmath8(d i ) at the 90% confidence limit&0.96&6600&281.7best @xmath8(d i)&1.16&5900&277.0maximum @xmath8(d i ) at the 90% confidence limit&1.45&3500&281.7maximum @xmath8(d i ) at the 99% confidence limit&1.61&2400&286.2 fig . [ extremes ] shows the observed deuterium profiles along with the expected absorption profiles ( upper and lower boundaries of the crosshatched regions ) whose shapes are determined by the shape of the n i profile ( heavy , solid line in fig . [ nando ] ) after it has been smoothed to allow for possible extra thermal doppler motions that would be expected for the lighter atoms ( dashed line in fig . [ nando ] ) . basically , apart from the thermal smearing , there is no evidence that there are deviations between the nitrogen and deuterium velocity profiles . in fig . [ extremes ] we also illustrate with a dashed line the depth and shape of the expected profile if @xmath96 were as high as @xmath97 , a value that would give @xmath98 as seen elsewhere ( [ bkgnd ] ) if @xmath99 ( [ n(hi ) ] below ) . when @xmath96 is forced to this large value , @xmath100 occurs at @xmath101k . this value seems unrealistically low , in view of the evidence that the 0@xmath211 rotational temperature of h@xmath38 toward @xmath1 ori a is 1625k ( savage et al . thus , we set a constraint @xmath102k ( but allowed other free parameters to float ) when we constructed the dashed line in fig . [ extremes ] . for this case , @xmath103 which is clearly unacceptable . published values of @xmath8 ( ) based on moderate resolution recordings of the l@xmath5 absorption in the spectrum of @xmath1 ori range from @xmath104 ( jenkins 1970 ) , based on photographic spectra recorded on sounding rocket flights , to @xmath105 ( bohlin , savage , & drake 1978 ) from a spectrum recorded by _ copernicus_. since @xmath1 ori is a hot star ( o9.5ii ) , the stellar absorption line makes a negligible contribution to @xmath8 ( ) . ori is 10@xmath106 @xmath107 . therefore correcting for the stellar l@xmath5 line changes log @xmath8(h i ) by only 0.01 dex . ] the accuracies of these measurements are satisfactory for studies of general trends , but for measuring d / h , and in particular to investigate the possible spatial variability of d / h , we must strive for a precision in @xmath8 ( ) that is as good as or preferably better than that for @xmath8 ( ) . the h absorption features shown in fig . [ dorideut ] are of no use in determining @xmath8 ( ) because the lines are heavily saturated , and most of the absorption is by small amounts of hydrogen with velocities well displaced from the line core . the damping wings of these lines are too weak to measure . in contrast , the l@xmath5 feature has very strong wings , ones that make this feature the least susceptible of all the lyman series lines to any contributions from high - velocity wisps of that do not produce detectable counterparts in absorption . it is for this reason that we concluded that the l@xmath5 feature was the best indicator of @xmath8(h i ) . spectra of @xmath1 ori obtained with the _ international ultraviolet explorer _ ( iue ) in high - dispersion mode ( fwhm @xmath108 25 km s@xmath109 ) were particularly attractive for our study of the l@xmath5 feature for several reasons . first , the alternative was to use archival _ copernicus _ data ( l@xmath5 was not recorded by imaps or hst ) , but the only available _ copernicus _ data with sufficiently broad wavelength coverage of l@xmath5 were obtained with the low - resolution u2 detector , and for this detector uncertainties are introduced by stray light from the vent hole ( rogerson et al . 1973 ) , an effect that requires a special correction ( bohlin 1975 ) of uncertain accuracy . second , a large number of observations of @xmath1 ori obtained under slightly different observing conditions ( e.g. , small aperture _ vs. _ large aperture ) over the course of many years are available from the iue archive . by analyzing all of the iue exposures rather than a single observation , we can validate our estimates of random errors and also increase our chances of exposing _ systematic _ errors in the derivation of @xmath8 ( ) . finally , our ability to combine many observations allowed us to reduce the effects from random noise by brute force . a potentially important source of systematic error is the fact that @xmath1 ori is a complex multiple - star system . the primary ( component a ) is a single - lined spectroscopic binary that is important in the history of ism research : stationary ca ii absorption features in its spectrum provided the earliest evidence of interstellar gas ( hartmann 1904 ) . the velocity amplitude of the binary is @xmath110 , and it has a period of 5.7 days ( harvey et al . it is also a partially eclipsing binary ( koch & hrivnak 1981 ) , and it has a visual companion of comparable brightness at a separation of @xmath111 ( heintz 1980 ; mcalister & hendry 1982 ) . the spectroscopic binary nature of the star system presents an opportunity to investigate systematic errors in the determination of the interstellar @xmath8 ( ) : the l@xmath5 profile is extremely broad , and if there are unrecognized stellar lines in the principal part of the lorentz wings of the l@xmath5 feature , then their additional optical depth could lead to an overestimate of @xmath8 ( ) . however , such stellar lines should move in velocity as the binary traverses its orbit , and this may lead to different values of @xmath8 ( ) when observations made at different times are analyzed . if this occurs , then we should see @xmath8 ( ) change as a function of the spectroscopic binary phase . similarly , we can check for systematic changes in @xmath8 ( ) when the multiple star enters the partial eclipses over the phase intervals [email protected] and [email protected] . while any dependence on phase may uncover an influence that stellar features have on the outcome , there is no guarantee that they do not perturb our result in a manner that is uniform over all phases . according to the nssdc archive , there are 59 iue observations of @xmath1 ori obtained with the short wavelength prime ( swp ) camera in the high - dispersion echelle mode . we screened these observations for saturated exposures , missing data , or other problems and rejected two of the observations , leaving 57 spectra for our analysis . using the standard iue rdaf software , we selected the spectral regions of interest from the standard iuesips data rather than the newsips reductions since there are a number of problems with newsips processing as applied to high dispersion spectra that could adversely affect our analysis ( massa et al . 1998 ) . in the vicinity of the l@xmath5 line the orders on the iue detector are closely spaced , and scattered light from adjacent orders overlaps in the interorder region causing an incorrect background subtraction and zero intensity level when using the standard software . we used the method of bianchi & bohlin ( 1984 ) to correct for this problem . also , in some cases , a velocity shift was applied to the iue data based on the position of the triplet at 1200 compared to that expected from the imaps profile ( [ template ] ) with optical depths rescaled to account for the stronger transition probabilities . this should register all of the iue data to the correct velocity scale to an accuracy of better than @xmath1125 km s@xmath109 , which is more than adequate for a determination of @xmath8 ( ) since we are fitting _ both _ of the strong damping wings of the l@xmath5 profile ( an error of 5 km s@xmath109 in the velocity scale zero point changes @xmath8 ( ) by an insignificant amount ) . iue data contain a number of perturbations in addition to the usual photon counting noise . they show strong fixed pattern noise , `` hot spots '' which mimic emission features , and drop - outs where the reseaux used to correct for camera distortions happen to fall on the spectrum ( harris & sonneborn 1987 ) . in addition , the spectra occasionally show artifacts at the transitions between echelle orders due to errors in the ripple correction ( see below ) . again , by analyzing _ all _ of the iue data , we can reduce the impact of these noise sources , which are present in some of the observations and are not apparent in others . we ascertained that there was no _ persistent _ nonlinearity in the photometric response of iue by comparing its l@xmath5 profiles of hd93521 and hd74455 with those recorded for the same stars by the ghrs on hst with the medium - resolution grating ( g160 m ) . departures from the ghrs spectra near the breaks in the iue echelle orders seem to come and go , but aside from the greater random noise in the iue spectra , the spectra are usually very similar to each other . in simplest terms , our means for constraining the column density followed a method introduced by jenkins ( 1971 ) and used later by bohlin ( 1975 ) , bohlin , et al ( 1978 ) , shull & van steenberg ( 1985 ) , and diplas & savage ( 1994 ) : we determined the @xmath8 ( ) that provides the best fit to the l@xmath5 profile with the optical depth @xmath113 at a given wavelength @xmath50 calculated from the expression @xmath114 ( jenkins 1971 ) , where @xmath115 is the l@xmath5 line center at the velocity centroid of the hydrogen . however , we went a step further by employing the technique used to estimate @xmath8 ( ) in [ n(di ) ] , i.e. , we first determined the important free parameters that could be adjusted to fit the l@xmath5 absorption profile , then we found the set of parameters that minimized @xmath116 using powell s method , and finally we set confidence limits on the column density by increasing ( or decreasing ) @xmath8 ( ) with the other parameters freely varying until @xmath116 increased by the appropriate amount for the confidence limit of interest . figure [ hicon ] shows a sample iue spectrum in the vicinity of the l@xmath5 absorption line . from this figure one can see that the continuum is close to linear in this region . however , it is possible that the continuum has a slight downward or upward curvature , so we assumed a second - order polynomial to describe the continuum and allowed the @xmath116 minimization process to determine the continuum shape that provided the best fit to the l@xmath5 profile . therefore the free parameters for fitting the profile were @xmath8 ( ) , three coefficients that specify the second - order continuum polynomial , and a simple additive correction to the intensity zero point . profile showed that the zero intensity level was not quite correct , so we included a zero point shift as a free parameter and used intensity points within the saturated core as one of the collections of terms for calculating @xmath116 . ] we point out that the continuum placement is constrained not only by the fits to regions that are far removed from the l@xmath5 feature , but also by requiring a good match to the shapes of its damping wings . for instance , if the continuum is badly placed or has too much upward or downward curvature , then a poor fit results . in particular , if we artificially forced the continuum to have a downward curvature ( in an experimental challenge to lower @xmath8(h i ) and thus provide a higher d / h ) , we obtained clearly inferior fits . we found that in the course of our minimizing @xmath3 that we could always simultaneously obtain a good fit to the profile and match the outlying fluxes with a nearly flat continuum . figure [ hiprofs ] shows two examples of l@xmath5 profiles observed with iue , along with the computed profiles for the lower and upper bounds on @xmath8 ( ) at the 90% confidence level for each case . the final continua corresponding to the upper and lower bounds are plotted with dashed lines . panel ( @xmath117 ) shows a typical spectrum , while ( @xmath118 ) shows examples of artifacts at @xmath119 1214.4 and 1224.8 that we encountered at the transitions between iue echelle orders . assuming these to be artifacts due to the ripple correction , we used only the sides of the l@xmath5 profile in the wavelength ranges [email protected] and [email protected] and thereby excluded these artifacts from the @xmath120 calculation . this procedure resulted in bounding profiles such as those shown in both of the panels of the figure . however , as an experiment we also processed all of the iue data including the region at @xmath121 1214.4 in the @xmath120 calculation in order to evaluate the importance of this effect on the final results ( see below ) . it is important to note that the great strength of the l@xmath5 profile makes the lorentzian wings dominate over the effects of instrumental or doppler broadening . gas that is known to exist in the vicinity of the orion association at high velocities ( @xmath122 ) ( cowie , songaila , & york 1979 ) should not be important , since the absorptions from any wisps of h i at such velocities are displaced by only about 0.3 relative to the line center . figure [ allhi ] shows the logarithms of the derived column densities with their 1@xmath92 error bars , plotted as a function of the spectroscopic binary phase , for all of the iue swp data except the two rejected exposures . we calculated the phase using the period and @xmath123 derived by harvey , et al . ( 1987 ) from their analysis of all suitable data from 1902@xmath211982 ( including some of the iue data used here ) . there are no obvious systematic trends as function of spectroscopic binary phase evident in this plot , which gives us some assurance that weak stellar lines do not significantly affect the derived @xmath8 ( ) . in the figure , large and small aperture data are shown with different symbols to check for any systematic differences , and no differences are readily apparent . with only the large aperture data we derive a mean column density of @xmath124()@xmath125 = 1.54@xmath126 @xmath107 with an rms dispersion @xmath92 = @xmath127 [ both quantities are weighted inversely by the variances of the individual @xmath8 ( ) estimates ] . with the small aperture data , we obtain @xmath124()@xmath125 = 1.59@xmath126 @xmath107 with @xmath92 = @xmath128 . therefore it appears appropriate to combine the large and small aperture data to constrain @xmath8 ( ) . using the entire iue data set , we obtain @xmath124()@xmath125 = 1.56@xmath126 @xmath107 with an rms scatter equal to @xmath129 . since there are 57 measurements , the error in the mean = @xmath130 = 0.01@xmath126 for the whole data set . we find that including the region that spans the ripple correction artifact at 1214.4 shown in figure [ hiprofs](b ) lowers the overall result by less than 0.01 dex . this is because the feature is present in only a small fraction of the iue spectra . the scatter in figure [ allhi ] appears to be due entirely to the uncertainties from noise in the individual measurements : the value for @xmath131/\sigma\left [ n({\rm h~i})\right]_i \right\}^2 $ ] calculated for the entire data set is 50.32 , which implies a reduced @xmath3 , @xmath132 . _ nevertheless , given the many potential sources of systematic error in these particular iue data , it is still possible that there are some unrecognized systematic errors which affect n(h i ) and that the real error in the mean is underestimated . _ however , the good fits to the l@xmath5 profiles ( see fig . [ hiprofs ] ) and the lack of pronounced variations with binary phase indicate that such unrecognized systematics are not likely to be large . absorption in the spectrum @xmath133 col , howk , savage and fabian ( 1999 ) estimated a systematic error of 0.02 dex for a determination of @xmath20 somewhat less than @xmath134 . this relative error is about half of the amount that would be needed to have any appreciable impact on the relative errors for our d / h toward @xmath1 ori a given in [ d / h ] . although we must acknowledge that their ghrs spectrum is of much better quality than any that were taken by iue , their independent estimate that gives a low value for the magnitude of a systematic error in this type of measurement is encouraging . ] it is reassuring to note that a constant , systematic error in the result for @xmath20 would need to be at least 15 times as large as the formal ( random ) error before it could have a meaningful effect on the overall error for d / h derived in [ d / h ] below . in the light of our deliberate attempt to uncover such systematic errors in the study of @xmath20 _ vs. _ orbital phase , we are confident that they could not have such a large numerical advantage , which means they are unlikely to be a critical issue in this investigation . finally , to illustrate more graphically our confidence in the h i result , we show with dotted and dash - dot damping profiles in fig . [ hiprofs](a ) the expected appearance of the l@xmath5 absorption if our @xmath8(d i ) derivations are correct but @xmath8(h i ) were low enough to make @xmath135 , as indicated for other lines of sight ( [ bkgnd ] ) . there seems to be no question that the real data are strikingly inconsistent with these lower values for @xmath8(h i ) . combining the results reported in [ n(di ) ] and [ n(hi ) ] , we find that with 90% confidence we can declare that @xmath136 in the direction of @xmath1 ori a. the most noteworthy feature of this result is that it differs from most determinations of d / h along other lines of sight in our local region of the galaxy ( [ bkgnd ] ) . it is clear that our value for d / h represents a deviation , even if one relies only on the hst observations that generally concentrate within the range @xmath137 in the very local medium and rejects the _ copernicus _ results because their accuracies may have been overstated . our result shows a simple velocity structure for the d i , n i and o i absorption profiles and thus removes the primary uncertainty that confronted laurent , et al ( 1979 ) . it also removes the grounds on which mccullough ( 1992 ) rejected the measurements of deuterium toward stars in orion . variations in the heavy element abundances of stars with similar ages , such as b stars in the orion association ( cunha & lambert 1994 ) or f and g type stars at a galactocentric radius nearly the same as the sun ( edvardsson et al . 1993 ) , suggest that the ism out of which the stars formed may be a heterogeneous mixture of gases with different levels of heavy element enrichment . this may possibly result from random dilutions of gas in the galactic plane by metal - poor material falling in from the halo ( meyer et al . if the material in the direction of @xmath1 ori had a lower deuterium abundance because it had been subjected to more intensive stellar processing or less of this dilution , we would expect the heavy element abundances to be higher than elsewhere . for our test of this proposition , we will examine the abundances of oxygen and nitrogen , both of which are unlikely to be appreciably altered by depletions onto dust grains ( meyer , cardelli , & sofia 1997 ; meyer , jura , & cardelli 1998 ) . o i and n i are also good standards because their ionizations are closely coupled to that of h i ( [ template ] ) , and this allows us to neglect the higher stages of ionization because they should be identified only with ionized h and d. o and n are also useful for comparisons with abundance studies elsewhere in the universe . these investigations are facilitated by the generous number of transitions in the ultraviolet with a broad range of @xmath49 values ( timmes et al . [ argon is another element that is expected to have very little depletion , but it is not suitable for comparison because under many circumstances its ionization can differ appreciably from that of h ( sofia & jenkins 1998 ) . ] meyer , et al . ( 1998 ) found from their measurement of the intersystem o i transition at 1356 that toward @xmath1 ori @xmath138 ( from the same data that we used to construct the profile shown in fig . [ nando ] ) . this value(h i ) adopted by meyer , et al . ( 1998 ) was @xmath139 which is very close to the result @xmath140 that we derived in our more precise analysis ( [ n(hi ) ] ) . ] is slightly less than their average of @xmath141 over 13 lines of sight . within the experimental errors , however , the value is consistent with the average . meyer , et al . ( 1997 ) found that @xmath142 toward 6 stars . since @xmath1 ori was not part of this sample , we must rely on our own measurement of n i which yields @xmath143 see [ template ] and the n i profile in fig . [ nando ] . with our result for @xmath8(h i ) , we arrive at @xmath144 . thus , we see no evidence that the gas has been specially enriched with material having a high abundance of heavy elements and , by virtue of a more intensive exposure to stellar interiors , a more thorough depletion of deuterium . in fact , the possible positive correspondence in the deviations in the n and d abundances is reminiscent of the suggested correlation ( but with large errors ) between d / h and zn / h shown by york & jura ( 1982 ) . of course , we must acknowledge that for nitrogen a comparison of our result for @xmath1 ori with the general measurements of meyer , et al ( 1997 ) for other lines of sight may be compromised by errors in f values , since we used different transitions to determine @xmath145 . our finding presented in [ d / h ] indicates that the most probable d / h toward @xmath1 ori is about half as large as that found from various hst investigations of the local ism . our result applies to an average over a range of velocities , which means that it represents a _ lower limit _ for the magnitude of deviations from the other cases . this anomaly is not linked with an increase in o / h or n / h , as one might expect from a simple explanation that a greater fraction of the gas had been cycled through stellar interiors . of course , it may be possible to envision that the gas toward @xmath1 ori holds an unusually large fraction of material that has cycled only through the _ outer envelopes _ of stars , thus depleting the d without increasing the concentrations of heavier elements . recent observations of hd emission in the infrared from gas near orion seem to confirm our finding that the abundance of deuterium is low in this region . wright , et al . ( 1999 ) detected emission from the hd @xmath146 transition at @xmath147 toward the orion bar with the long wavelength spectrometer on board the _ infrared space observatory _ ( iso ) . while there are some uncertainties in the rotation temperature of hd and the correction factors that must be applied to observations of the accompanying h@xmath38 , they arrived at a preferred value @xmath148 which leads to @xmath149 since there should be no appreciable d or h in atomic form . their total range for d / h could be as large as 0.35 to @xmath150 however . bertoldi , et al . ( 1999 ) used the short wavelength spectrometer on iso and found a weak emission from the @xmath151 transition at @xmath152 for hd in the orion molecular outflow omc-1 . again , corrections using information from models of the gas had to be made to interpret the results . bertoldi , et al.(1999 ) concluded that @xmath153 . the measurements of hd by both groups seem to lead to results that are consistent with our determination of atomic d / h toward @xmath1 ori a. very distant gas systems that are registered in quasar absorption line spectra reveal apparent values of d / h that range from @xmath154 ( burles & tytler 1998a , b ) to @xmath155 ( songaila et al . 1994 ; carswell et al . 1996 ; rugers & hogan 1996 ; wampler et al . 1996 ; webb et al . 1997 ; tytler et al . 1999 ) see reviews by burles & tytler ( 1998c ) and hogan ( 1998 ) . the large dispersion in these outcomes might be attributable either to complications that arise from our incomplete knowledge of the chemical evolution of systems at large redshifts , the difficulty in obtaining accurate values of @xmath8(h i ) , or to the presence of random , weak , h i systems that masquerade as deuterium by having a velocity offset that is equal to about @xmath156 from a main system . one might suppose that , in time , additional observations that include new cases or new data on existing ones may lead to a better understanding of the behavior of d / h in the universe . unfortunately , this optimistic belief may have been dealt a setback by our result that indicates that d / h could be driven by a process that we do not understand . essentially , we see evidence that the ratio changes over a distance scale where the environment should be homogeneous , according to observations of other elements and generally accepted simple models for a galaxy s chemical evolution and mixing rates in the ism . a few proposals to explain possible deviations in the balance of atomic d to h in the interstellar medium have been considered in the past . the simplest involves the selective incorporation of d into hd , an effect that can amplify hd / h@xmath38 to values well above the fundamental ratio of d to h ( watson 1973 ) , but one that is probably counterbalanced by the more rapid photodissociation of hd in diffuse clouds because there is no self shielding ( as there often is with h@xmath38 ) . a preferential formation of hd is not responsible for the depletion of atomic d toward @xmath1 ori a , since @xmath157 ( spitzer , cochran , & hirshfeld 1974 ) ( likewise , we see no hd features in our imaps spectrum ) . another alternative , one advanced by vidal - madjar , et al . ( 1978 ) and bruston et al . ( 1981 ) , makes use of the differences in the ways that d and h can respond to radiation pressure , as a result of the very different opacities in the lyman lines . they proposed that this effect could lead to a separation of the two species if there were a density gradient and a nonisotropic radiation field . finally , jura ( 1982 ) has suggested that deuterium atoms could collide with dust grains , stick to them , and then be more strongly bound than hydrogen . furthermore , he suggests that the mobility of the d atoms on the surfaces of these grains could be much lower than that of h , thus limiting the chances for combining with h atoms and being ejected as hd . one possible way to investigate the plausibility of this hypothesis might be to look for d@xmath21c or d@xmath21o stretch mode absorption features in dense clouds . we look forward to the possibility that insights on the relationship for the variability in d / h to other interstellar parameters could arise from the anticipated large increase of information that should come from the _ far ultraviolet spectroscopic explorer _ ( fuse ) after its launch in mid-1999 . for the local ism , the fuse principal investigator team has identified as potential targets 7 cool stars , 19 white dwarfs , 9 late b- or early a - type stars , and 7 central stars of planetary nebulae for studying d / h in the first two years of operations . the orfeus - spas project was a joint undertaking of the us and german space agencies , nasa and dara . the successful execution of our observations was the product of efforts over many years by engineering teams at princeton university observatory , ball aerospace systems group ( the industrial subcontractor for the imaps instrument ) and daimler - benz aerospace ( the german firm that built the astro - spas spacecraft and conducted mission operations ) . contributions to the success of imaps also came from the generous efforts by many members of the optics branch of the nasa goddard space flight center ( grating coatings and testing ) and from o. h. w. siegmund and s. r. jelinsky at the berkeley space sciences laboratory ( deposition of the photocathode material ) . this research was supported by nasa grants nag5@xmath21616 to princeton university and nag5@xmath213539 to villanova university . we thank k. r. sembach for supplying an idl routine to apply the bianchi & bohlin correction to the iue data . we also thank a. vidal - madjar , b. t. draine and b. d. savage for their helpful comments on early drafts of this paper . the o i absorption feature at 1355 was observed by the nasa / esa hubble space telescope . this spectral segment was obtained from the data archive at the space telescope science institute , operated by aura under nasa contract nas5 - 26555 . the iue data were obtained from the national space science data center ( nssdc ) at nasa s goddard space flight center . glz , m. , kappelmann , n. , appenzeller , i. , barnstedt , j. , fromm , a. , grewing , m. , gringel , w. , haas , c. , hopfensitz , w. , krmer , g. , krautter , j. , lindenberger , a. , mandel , h. , werner , k. , & widmann , h. 1998 , in proc . iau colloq . 166 , the local bubble and beyond , ed . d. breitschwerdt , m. j. freyberg , & j. trmper ( berlin : springer ) , p. 75 harris , a. w. , & sonneborn , g. 1987 , in exploring the universe with the iue satellite , ed . y. kondo , w. wamsteker , a. boggess , m. grewing , c. de jager , a. l. lane , j. l. linsky , & r. wilson ( dordrecht : reidel ) , p. 729 jenkins , e. b. , joseph , c. l. , long , d. , zucchino , p. m. , carruthers , g. r. , bottema , m. , & delamere , w. a. 1988 , in ultraviolet technology ii , ed . r. e. huffman ( bellingham : the international society for optical engineering ) , p. 213	studies of the abundances of deuterium in different astrophysical sites are of fundamental importance to answering the question about how much deuterium was produced during big bang nucleosynthesis and what fraction of it was destroyed later . with this in mind , we used the interstellar medium absorption profile spectrograph ( imaps ) on the orfeus - spas ii mission to observe at a wavelength resolution of @xmath0 ( fwhm ) the l@xmath1 and l@xmath2 absorption features produced by interstellar atomic deuterium in the spectrum of @xmath1 ori a. a @xmath3 analysis indicated that @xmath4 at a 90% level of confidence , and the gas is at a temperature of about 6000k . in deriving these results , we created a template for the velocity profile defined by 7 different n i transitions recorded at a high signal - to - noise ratio . extra free parameters in the analysis allowed for the additional uncertainties that could arise from various sources of systematic error . to derive a value for d / h , we measured the l@xmath5 absorption features in 57 spectra of @xmath1 ori in the iue archive , with the objective of arriving at a more accurate h i column density than those reported by other investigators . from our measurement of @xmath6 , we found that @xmath7 ( 90% confidence ) . systematic errors in the derivation of @xmath8(h i ) probably dominate over the very small formal error , but their relative value should be smaller than that for @xmath8(d i ) . our result for d / h contrasts with the more general finding along other lines of sight that @xmath9 . the underabundance of d toward @xmath1 ori a is not accompanied by an overabundance of n or o relative to h , as one might expect if the gas were subjected to more stellar processing than usual .
since the discovery of a new phase of matter in @xmath0 = 200 agev heavy ion collisions @xcite , the goal of the high energy nuclear physics community is now the quantitative determination of the properties of the quark - gluon plasma ( qgp ) . for instance , is the medium produced at rhic and lhc a weakly - coupled gas of the quarks and gluons of the quantum chromodynamics ( qcd ) lagrangian or perhaps a strongly coupled fluid made up of emergent collective degrees of freedom , and how does this description change as a function of @xmath0 , @xmath5 , and centrality ? high momentum particles provide the most direct probe of the soft momentum modes present in the qgp created in heavy ion collisions ( see @xcite and references therein ) . extracting information about the fundamental properties of the qgp requires a comparison between theoretical calculations and experimental data . despite its great promise , to date no one energy loss model has provided a consistent picture of the different relevant observables , such as and , as a function of particle mass , @xmath6 , centrality , and @xmath0 . we demonstrate here that calculations based on the whdg energy loss model @xcite which is itself based on weak - coupling , perturbative qcd ( pqcd ) methods provide a reasonably good , qualitatively consistent description of the preliminary data on the suppression and azimuthal anisotropy of light hadrons and the suppression of @xmath2 mesons as a function of @xmath6 and centrality at lhc when constrained by a rigorous statistical analysis of central rhic @xmath1 suppression data @xcite . on the other hand , calculations based on the strong - coupling techniques in an ads / cft heavy quark drag model appear to significantly overpredict the central values of the preliminary @xmath2 meson suppression when qualitatively constrained to the non - photonic electron data from rhic @xcite . however , the uncertainties related to this heavy meson measurement are quite large , and currently the strong - coupling calculation does not appear to be falsified . a generic energy loss calculation requires a convolution of a hard production spectrum followed by the in - medium energy loss and a fragmentation function . in whdg @xcite , the parton is allowed to lose energy perturbatively through both elastic and inelastic processes in a realistically described , 1d bjorken - expanding medium . the radiative energy loss equations were found via the use of standard feynman diagram techniques @xcite , making sure to take into account the crucial fact that the parton is created at a finite time . it is important to note that even at top lhc energies , we find elastic energy loss @xcite provides a sizable ( @xmath725% ) fraction of the total energy loss @xcite . using thermal field theory to relate density to temperature and temperature to the debye mass and mean free path in a qgp , in whdg the only free parameter becomes the medium gluon density @xmath8 ( with @xmath9 fixed and geometry set by the colliding nuclei ) . as in @xcite , in this work we assume that the medium density scales precisely with the measured charged particle multiplicity , @xmath10 . we use the rigorous statistical comparison @xcite of whdg predictions as a function of @xmath8 at @xmath11 agev to rhic data to set the proportionality constant between @xmath8 and @xmath10 . with the published @xmath10 data from alice @xcite , and continuing to keep @xmath9 fixed , we can compute all observables at lhc _ using no free parameters_. a sample of such zero parameter predictions is shown in and . the prediction associated with the best fit value of the medium density is denoted by a dashed line ; the one - sigma uncertainty associated with that best fit parameter extraction is denoted by a band . in ( a ) we demonstrate a good qualitative agreement of the normalization and momentum dependence between the preliminary cms @xmath12 data @xcite and @xmath13 predictions from whdg @xcite ; quantitatively , the calculations overpredict the suppression by , at worst , somewhat more than a standard deviation . in ( b ) and ( c ) we show that the whdg calculations are in excellent quantitative agreement with the preliminary atlas measurements @xcite at several centrality bins . in ( a ) we again see excellent agreement between the whdg predictions of @xmath2 meson suppression and the preliminary alice data @xcite within the stated uncertainties . we provide for future reference whdg predictions at @xmath14 atev of @xmath3 meson suppression in ( b ) and also of the double ratio of @xmath15 in ( c ) . note the rapid convergence of the pqcd - based predictions to unity in ( c ) . @xmath16{cmsraab.eps } & \includegraphics[width=2.1in]{pi0v22763040watlasb.eps } & \includegraphics[width=2.1in]{pi0v22765060watlasb.eps}\\[-.1 in ] \qquad\mbox{\footnotesize(a ) } & \qquad\mbox{\footnotesize(b ) } & \qquad\mbox{\footnotesize(c ) } \end{array}$ ] in the ads / cft drag model used here the feynman diagrammatic approach in 4d is replaced by a classical string derivation in 5d @xcite . it was seen in @xcite that these strong coupling techniques yield a model whose suppression of non - photonic electrons at @xmath11 agev is in both qualitative and quantitative agreement with the data taken at rhic . again assuming that the medium density scales with observed charged particle multiplicity and keeping all other parameters the same produces the ads / cft drag curves shown in ( a)-(c ) . the three ads / cft drag curves shown explore part of the possible parameter space associated with linking qcd parameters to those in @xmath17 sym @xcite . clearly the drag predictions quantitatively overpredict the suppression of @xmath2 mesons compared to the preliminary alice data @xcite ; however , due to the large uncertainties in that data , the theoretical calculation can not currently be claimed to be falsified . future distinguishing measurements include very different -dependencies for @xmath3 meson suppression and for the double ratio of @xmath18 . both of these quantities increase as a function of momentum for pqcd - based calculations . on the other hand , ads / cft drag - based calculations robustly predict a decrease of @xmath19 as a function of and a near -independence of the double ratio of @xmath18 . one must note that these drag calculations are performed in @xmath17 sym , not qcd , and in the limit that @xmath20 and large @xmath21 . additionally the drag calculations assume that the quark mass is very large compared to the temperature of the plasma , an assumption that may not hold particularly well for the charm quark . finally , there is a known upper - bound to the velocity of the heavy quark for which the ads / cft derivations are applicable @xcite . this upper limit is mass- , @xmath21- , and temperature - dependent . we give an indication of where corrections to the lowest - order ads / cft calculations may set in ( at the highest temperature , @xmath22 , where @xmath23 is the thermalization time ) with an `` ( '' on the relevant curves in ( a)-(c ) and where corrections almost certainly set ( at the lowest temperature @xmath24 mev ) in with an `` ] '' . @xmath25{dmesonsi.eps } & \includegraphics[width=1.8in]{braab.eps } & \includegraphics[width=1.8in]{dtobraae.eps } \\[-.1 in ] \qquad\mbox{\footnotesize(a ) } & \qquad\mbox{\footnotesize(b ) } & \qquad\mbox{\footnotesize(c ) } \end{array}$ ] at rhic kinematics , no energy loss calculation provided a consistent picture of the multiple observables reported @xcite . in particular , constraining pqcd - based calculations to @xmath1 suppression led to a qualitative underprediction of the @xmath26 gev / c@xmath27 @xcite and a quantitative underprediction of non - photonic electron suppression @xcite . however , constrained to rhic data , we find the pqcd - based whdg energy loss calculation qualitatively describes the light hadron suppression and quantitatively describes the azimuthal anisotropy and heavy meson suppression at lhc , using _ no free parameters_. nevertheless , one must be cautious in these comparisons to data as there are a large number of theoretical uncertainties that are not yet taken into account in the whdg nor fully in any other pqcd - based energy loss model : higher order effects in opacity @xcite , coupling @xcite , heavy quark mass ( divided by parton energy ) @xcite , collinearity @xcite ; initial conditions @xcite ; energy loss in confined matter @xcite . on the other hand , the heavy quark drag calculations appear to oversuppress the @xmath2 mesons compared to the central values of the data , although the results are not completely inconsistent with the data within the current , large experimental uncertainties . similar to the perturbative calculations , there are many theoretical uncertainties in the drag calculations that should give one pause when comparing to experimental data ; for instance the derivations were performed in @xmath17 sym @xcite and higher order corrections from , e.g. , loops and quark mass to plasma temperature @xcite have not yet been taken in to account . we look forward to the future measurements of @xmath3 meson suppression and the double ratio of @xmath18 at lhc which will help distinguish between the energy loss mechanics dominant in heavy ion collisions .	we compare calculations of jet quenching observables at @xmath0 = 2.76 atev to preliminary lhc data from weak - coupling pqcd and strong - coupling ads / cft drag energy loss models . rigorously constrained to @xmath0 = 200 agev rhic @xmath1 suppression data and introducing no free parameters , the pqcd - based whdg model simultaneously describes qualitatively the suppression of light hadrons and quantitatively the azimuthal anisotropy ( over many centrality classes ) of light hadrons and the suppression of @xmath2 mesons at lhc . the drag predictions shown here qualitatively constrained to rhic data compare poorly to the measured suppression of @xmath2 mesons , but the current experimental uncertainties are large . the double ratio of @xmath2 to @xmath3 meson @xmath4 should provide a robust experimental measurement to distinguish between the very different assumptions of the strength of interactions in the qgp produced in heavy ion collisions ; i.e. , whether , from a jet quenching standpoint , the medium is either weakly- or strongly - coupled . address = department of physics , university of cape town , private bag x3 , rondebosch 7701 , south africa
following previous papers @xcite in which we computed non - perturbatively the renormalization constants of composite operators with overlap fermions in quenched qcd , this paper will study the scaling behavior of the renormalization constants . we adopt the non - perturbative renormalization method which was introduced by martinelli @xmath0 @xcite . the method allows a full non - perturbative computation of the matrix elements of composite operators in the regularization independent ( ri ) scheme @xcite ( it is called the @xmath1 scheme by chetyrkin @xcite ) . the matching between the @xmath2 scheme and @xmath3 , which is intrinsically perturbative , is computed using continuum perturbation theory , which is well behaved . the overlap fermion @xcite was proposed by narayanan and neuberger to evade the so called no - go " theorem @xcite . the action in the massless limit preserves a lattice form of chiral symmetry even at finite lattice spacing and volume @xcite . the use of the overlap action entails many theoretical advantages : it has no additive mass renormalization , there are no order @xmath4 artifacts , and it has very good scaling @xcite . the renormalized operator @xmath5 is related to the bare operator , @xmath6 , calculated on the lattice via @xmath7 in this work , we will consider the fermion operators @xmath8 where @xmath9 are the dirac gamma matrices @xmath10 and the corresponding notations will be \ { s , v , p , a , t } respectively . the renormalization condition is imposed directly on the three - point vertex function @xmath11 , which is calculated in a fixed gauge , _ e.g. _ , landau gauge in our case , at a momentum scale @xmath12 @xmath13 here @xmath14 is the field or wave - function renormalization constant , @xmath15 . it is apparent that we can only get the ratio of the renormalization constant @xmath16 for the operator @xmath17 and the wave - function renormalization constant @xmath18 , from the renormalization condition of eq . ( [ eq : ren_con ] ) . in order to obtain the renormalization constant @xmath16 for the operator @xmath17 , one needs to know @xmath14 first . in this work , we will obtain @xmath14 directly from the quark propagator . it can be defined from the ward identity ( wi ) as @xcite @xmath19 which , in landau gauge , differs from @xmath18 by a finite term of order @xmath20 . the matching coefficients have been computed using continuum perturbation theory @xcite . we work on three lattices , each with a different lattice spacing , @xmath4 , but having similar physical volume . lattice parameters are summarized in table 1 . the quenched gauge configurations are created using a tadpole improved plaquette plus rectangle ( lscher - weisz ) gauge action and gauge fixed to the landau gauge using a conjugate gradient fourier acceleration algorithm . [ parameter ] the overlap - dirac operator we use is @xmath21 where @xmath22 is regulator mass and @xmath23 is bare quark mass , and @xmath24 is the matrix sign function of an hermitian operator @xmath25 = @xmath26 . we use the tadple improved wilson kernel @xmath27 in the overlap operator , and @xmath28 is used for the regulator mass @xmath22 for all three lattices . we calculate the overlap quark propagator for 15 bare quark masses on three lattices , they are @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33,@xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , and @xmath43 mev respectively our calculation begins with the evaluation of the inverse of the overlap - dirac operator . after we calculate the quark propagator in coordinate space for each configuration , we use the landau gauge fixing transformation matrix to rotate the quark propagator to landau gauge . then the discrete fourier transformation is used to obtain the quark propagator in momentum space . afterward , we calculate five projected vertex functions @xmath11 for each bare quark mass , and extrapolate to the chiral limit . these projected vertex functions @xmath11 are in general dependent on @xmath44 . the dependence may come from two sources . one is from the usual running of the renormalization constant in the @xmath2 scheme . the other is from possible @xmath44 errors and we need to remove this . in order to confront experiment , it is preferable to quote the final results in the @xmath45 scheme at a certain scale . one needs to transform the results in the @xmath2 scheme to the @xmath45 scheme . the detailed analysis can be found in refs . we work on three lattices with similar physical volumes and different spacings @xmath4 to investigate the the scaling behavior of the renormalization constants . here we compare the results of renormalization constants @xmath18 , @xmath46 , @xmath47 and @xmath48 on the different lattices in the @xmath49 scheme at 2.0 gev . fig . [ fig3lz ] shows the four renormalization constants @xmath18 , @xmath46 , @xmath47 and @xmath48 against the square of the lattice spacing @xmath4 . because overlap fermions are free of @xmath50 errors , the leading term must be proportional to @xmath51 . we use a simple linear fit @xmath16 = @xmath52 + @xmath53 @xmath51 , and take @xmath52 as the value of the renormalization constant @xmath16 in the continuum limit . the numerical values of calculated renormalization constants in the continuum limit in the @xmath49 and ri schemes at 2.0 gev are displayed in table 2 . in the continuum , the vector current is conserved , so @xmath46 should be equal to one . for the axial vector current , due to the pcac relation , it will be conserved at large momenta , where the anomaly has no effect . our result for @xmath46 and @xmath54 in table 2 compares favorably with 1 . j. b. zhang , d. b. leinweber , k. f. liu , and a. g. williams nucl . ( proc . suppl . ) b 128 , 240 ( 2004 ) . j. b. zhang , n. mathur , s. j. dong , t. draper , i. horvath , f. x. lee , d. b. leinweber , k.f . liu , and a. g. williams . [ arxiv : hep - lat/0507022 ] .	we compute non - perturbatively the renormalization constants of composite operators for overlap fermions by using the regularization independent scheme . the scaling behavior of the renormalization constants is investigated using the data from three lattices with similar physical volumes and different lattice spacings . the approach of the renormalization constants to the continuum limit is explored .
this paper revolves around the excess degree of a @xmath1-dimensional polytope @xmath2 , or simply @xmath1-polytope , and some of its applications . we define the _ excess degree _ @xmath0 , or simply excess , of a @xmath1-polytope @xmath2 as the sum of the excess degrees of its vertices , i.e. @xmath14 here as usual @xmath15 denotes the _ degree _ of a vertex @xmath16 , i.e the number of edges of @xmath2 incident with the vertex ; @xmath17 denotes the set of vertices of @xmath2 ; and @xmath4 and @xmath5 denote the number of vertices and edges of the polytope . our first substantial result , in 3 , is the excess theorem : the smallest values of the excess degree of @xmath1-polytopes are 0 and @xmath7 ; clearly a polytope is simple if , and only if , its excess degree is 0 . note that for fixed @xmath1 and @xmath4 , the possible values of the excess are either all even or all odd . we further show that if @xmath1 is even , the excess degree takes every even natural number from @xmath18 onwards ; while , if @xmath1 is odd , the excess degree takes every natural number from @xmath19 onwards . so , for a fixed @xmath1 , only a finite number of gaps in the values of the excess of a @xmath1-polytope are possible . in 4 , we study @xmath1-polytopes with excess strictly less than @xmath1 , and establish some similarities with simple polytopes . in particular , they are either pyramids or decomposable . ( a polytope @xmath2 is _ ( minkowski ) decomposable _ if it can be written as the minkowski sum of two polytopes , neither of which are homothetic to @xmath2 ; otherwise it is _ indecomposable_. ) a polytope can have excess @xmath8 only if @xmath9 or 5 , which shows that there are infinitely many gaps in the possible values of the excess degree . polytopes with excess @xmath7 exist in all dimensions , but their structure is very restricted : either there is a unique nonsimple vertex , or there is a @xmath20-face containing only vertices with excess degree @xmath21 . a similar , but slightly more complicated , situation arises for excess @xmath8 : a 5-polytope with excess 4 may also contain 2 vertices with excess degree 2 . in 5 , we characterise all the decomposable @xmath1-polytopes with @xmath12 or fewer vertices ; this incidentally proves that a conditionally decomposable @xmath1-polytope must have at least @xmath22 vertices . the final application , in 7 , is the completion of the @xmath13 table for @xmath23 ; that is , we give all the possible values of @xmath13 for which there exists a @xmath1-polytope with @xmath24 , @xmath4 vertices and @xmath5 edges . the solution of this problem for @xmath25 was already well known ( * ? ? ? * chap . the same result has recently been independently obtained by kusunoki and murai @xcite . our proof requires some results of independent interest ; in particular , a characterisation of the 4-polytopes with 10 vertices and minimum number of edges ( namely , 21 ) ; this is completed in 6 . we have more comprehensive results characterising polytopes with a given number of vertices and minimum possible number of edges , details of which will appear elsewhere @xcite . most of our results tacitly assume that the dimension @xmath1 is at least 3 . when @xmath26 , all polytopes are both simple and simplicial , and the reader can easily see which theorems remain valid in this case . in this subsection we group a number of basic results on polytopes which we will use throughout the paper . recall that a _ facet _ of a @xmath1-polytope is a face of dimension @xmath8 ; a _ ridge _ is a face of dimension @xmath7 ; and a _ subridge _ is a face of dimension @xmath27 . a fundamental property of polytopes is that every ridge is contained in precisely two facets . recall that a vertex @xmath16 in a @xmath1-polytope @xmath2 is called _ simple _ if its degree in @xmath2 is precisely @xmath1 ; equivalently if it is contained in exactly @xmath1 facets . otherwise it is _ nonsimple_. note that a nonsimple vertex in a polytope @xmath2 may be simple in a proper face of @xmath2 ; we often need to make this distinction . a polytope is _ simple _ if every vertex is simple ; otherwise it is _ nonsimple_. we will call @xmath28 the * underfacet * corresponding to @xmath29 . klee @xcite called this the face figure , but that term is sometimes used for a different concept @xcite . when @xmath29 has dimension 0 , the underfacet is simply the vertex figure . let @xmath30 be a hyperplane intersecting the interior of @xmath2 and containing no vertex of @xmath2 , and let @xmath31 be one of the two closed half - spaces bounded by @xmath30 . set @xmath32 . if the vertices of @xmath2 not in @xmath31 are the vertices of a face @xmath29 , then the polytope @xmath33 is said to be obtained by _ truncation _ of the face @xmath29 by @xmath30 . we often say that @xmath2 has been _ sliced _ or _ cut _ at @xmath29 . [ lem : face - all - neighbours - simple ] let @xmath2 be a @xmath1-polytope and let @xmath29 be a face of @xmath2 . suppose the neighbours outside @xmath29 of every vertex of @xmath29 are all simple in @xmath2 . let @xmath30 be a hyperplane as above , and let @xmath32 be obtained by truncation of @xmath29 by @xmath30 . then every vertex in the facet @xmath28 is simple in @xmath33 , and thus the facet @xmath28 is a simple @xmath34-polytope . every vertex @xmath35 in the facet @xmath28 is the intersection of @xmath30 and an edge @xmath36 of @xmath2 outside @xmath29 but incident to a vertex of @xmath29 . consequently , the facets of @xmath33 containing the vertex @xmath35 in @xmath28 are precisely the facets of @xmath2 containing the edge @xmath36 plus the facet @xmath28 . since the edge @xmath36 is contained in exactly @xmath8 facets of @xmath2 , the vertex @xmath35 in @xmath33 is contained in exactly @xmath1 facets of @xmath33 , thus is simple in @xmath33 . an interesting corollary of [ lem : face - all - neighbours - simple ] arises when @xmath29 is a vertex and reads as follows . [ lem : vertex - all - neighbours - simple ] let @xmath2 be a @xmath1-polytope and let @xmath37 be a vertex whose neighbours are all simple in @xmath2 . then the vertex figure of @xmath2 at @xmath37 is a simple @xmath34-polytope . the next few remarks are simple but useful . [ rmk : two - vertices ] for any edge of a polytope , and any vertex outside that edge , there is a facet containing the edge but not the vertex . hence for any two vertices of a polytope , there is a facet containing one of the vertices but not the other . [ rmk : simplenonsimplevertex ] let @xmath2 be a @xmath1-polytope and let @xmath37 be a vertex simple in a facet @xmath29 . suppose that every facet containing @xmath37 intersects @xmath29 at a ridge . then @xmath37 is a simple vertex in @xmath2 . the other facets containing @xmath37 are given by the ridges of @xmath29 containing @xmath37 , whose number is @xmath8 since @xmath37 is simple in @xmath29 . thus , in total there are @xmath1 facets containing @xmath37 . [ rmk : simplevertexoutside ] let @xmath2 be a polytope , @xmath29 a facet of @xmath2 and @xmath16 a nonsimple vertex of @xmath2 which is contained in @xmath29 . if @xmath16 is adjacent to a simple vertex @xmath38 of @xmath2 in @xmath39 , then @xmath16 must be adjacent to another vertex in @xmath39 , other than @xmath38 . since @xmath16 is nonsimple , it is contained in at least @xmath40 facets of @xmath2 . the edge @xmath41 is contained in exactly @xmath8 facets of @xmath2 , since @xmath38 is simple . hence there are at least two edges of @xmath16 outside @xmath29 , as desired . [ lem : non - disjointfacets ] let @xmath29 and @xmath42 be any two distinct nondisjoint facets of a @xmath1-polytope @xmath2 and let @xmath43 . a. every vertex in @xmath44 has excess degree at least @xmath45 . b. the total excess degree of @xmath2 is at least @xmath46 . c. if @xmath44 is not a ridge , then @xmath2 is not simple . set @xmath47 . then , for any vertex @xmath16 in @xmath48 , the degrees of @xmath16 in @xmath48 , @xmath49 and @xmath50 are at least @xmath51 , @xmath52 and @xmath52 , respectively . so the total degree of @xmath16 in @xmath53 is at least @xmath54 . this implies that the excess degree of each vertex in @xmath48 is at least @xmath55 . since there are at least @xmath56 vertices in @xmath48 , the total excess degree of @xmath2 is at least @xmath57 . in particular , if two facets of a @xmath1-polytope intersect in a face @xmath58 with @xmath59 , then every vertex in @xmath58 is nonsimple in the polytope . we will call a polytope * semisimple * if every pair of facets is either disjoint , or intersects in a ridge . clearly every simple polytope is semisimple . [ rmk : facets - intersecting - at - ridges ] let @xmath2 be a semisimple @xmath1-polytope . then every facet containing a nonsimple vertex of @xmath2 is also nonsimple . furthermore , each nonsimple vertex of @xmath2 is nonsimple in each of the facets containing it . if @xmath9 , then [ rmk : simplenonsimplevertex ] gives at once that the polytope is simple . so we let @xmath60 . let @xmath16 be a nonsimple vertex of @xmath2 and let @xmath29 be some facet containing it . since every other facet containing @xmath16 intersects @xmath29 at a ridge , there are at least @xmath1 ridges of @xmath29 containing @xmath16 , which implies that @xmath16 is nonsimple in @xmath29 , and @xmath29 is therefore nonsimple . in this subsection we introduce or recall families of polytopes which are important for this work . there is a 3-polytope with six vertices , ten edges and six facets ( four quadrilaterals and two triangles ) , which can fairly be described as the simplest polyhedron with no name . accordingly , we will call it @xmath61 ; see [ fig : polytopes](d ) . this humble example will naturally appear several times in this paper . ( kirkman @xcite called it a 2-ple zoneless monaxine heteroid , having an amphigrammic axis , but this terminology never caught on . ) call _ simplicial @xmath1-prism _ any prism whose base is a @xmath34-simplex . we will often refer to these simply as prisms . they each have @xmath62 vertices , @xmath63 edges , and @xmath64 facets . being simple , they have excess degree 0 . @xmath65-triplex _ as a @xmath66-fold pyramid over the simplicial @xmath67-prism , @xmath68 , and denote it by @xmath69 . any triplex is of this form for some values of @xmath1 and @xmath67 , and clearly has @xmath70 vertices . clearly @xmath71 is a simplex , and @xmath72 is a simplicial @xmath1-prism . it is well known ( see @xcite and ( * ? ? ? * ch . 10 ) ) and easily checked , that triplices ( other than the simplex ) have @xmath64 facets . they were studied in some detail in @xcite . it is routine to check that @xmath69 has excess degree @xmath73 . more generally , denote by @xmath74 the sum of an @xmath75-dimensional simplex and an @xmath76-dimensional simplex lying in complementary subspaces . the polytope @xmath74 is a simple @xmath77-dimensional polytope with @xmath78 vertices , @xmath79 edges and @xmath80 facets . likewise @xmath81 will denote the sum of three simplices , @xmath75-dimensional , @xmath76-dimensional and an @xmath82-dimensional respectively , lying in complementary subspaces . [ rmk : dplus2facets ] any polytope with @xmath64 facets is combinatorially isomorphic to an @xmath83-fold pyramid over @xmath74 for some values of @xmath84 ; see ( * ? ? ? * sec 6.1 ) and @xcite . in particular , a simple polytope with @xmath64 facets must be @xmath74 for some @xmath85 with @xmath86 . the @xmath1-dimensional _ pentasm _ , or just @xmath1-pentasm , was defined in @xcite as the minkowski sum of a simplex and a line segment which is parallel to one triangular face , but not parallel to any edge , of the simplex ; or any polytope combinatorially equivalent to it . the same polytope is obtained by truncating a simple vertex of the triplex @xmath87 . pentasms have @xmath12 vertices and excess degree @xmath7 ; see [ fig : polytopes](a)-(b ) for drawings of them . [ rmk : pentasm - facets]the facets of @xmath1-pentasm , @xmath88 in total , are as follows . * @xmath7 copies of a @xmath34-pentasm , * 2 simplicial prisms , and * 3 simplices . we can consider the pentagon as a 2-dimensional pentasm . take any polytope @xmath2 with dimension @xmath89 and stack a vertex @xmath90 beyond one facet and move it slightly so that @xmath37 is contained in the affine hull of @xmath91 other facets ( @xmath92 ) . by _ stacking _ we mean as usual adding a vertex beyond a facet of @xmath2 and taking the convex hull . a point @xmath93 is _ beyond _ a face @xmath48 of @xmath2 if the facets of @xmath2 containing @xmath48 are precisely those that are visible from @xmath93 , where a facet @xmath29 of @xmath2 is said to be _ visible _ from the point @xmath93 if @xmath93 belongs to the open half - space determined by the affine hull of @xmath29 which does not meet @xmath2 . all the @xmath7 polytopes constructed in this manner have the same graph , some have even the same @xmath20-skeleton . if @xmath2 is taken to be a simplicial @xmath1-prism , we call the resulting polytopes _ capped @xmath1-prisms_. consider a capped @xmath1-prism . denote by @xmath67 the minimum dimension of any face of the simplicial @xmath1-prism whose affine hull contains the _ extra _ vertex @xmath90 . the combinatorial type of such a polytope depends on the value of @xmath67 ; let us denote it by @xmath94 . note that @xmath95 , otherwise @xmath90 would be a vertex of the prism . for @xmath96 , the capped prism @xmath97 will be ( combinatorially ) just another prism . for @xmath98 , @xmath99 is a pentasm , with @xmath100 edges . for @xmath101 and fixed @xmath89 , the @xmath7 polytopes @xmath94 are combinatorially distinct , although their graphs are all isomorphic , with @xmath12 vertices and @xmath102 edges . in particular , they all have excess degree @xmath1 . however @xmath94 has @xmath103 facets , so their @xmath104-vectors are all distinct . denote by @xmath105 a polytope obtained by slicing a one nonsimple vertex of a @xmath106-triplex . this polytope can be also realised as a prism over a @xmath20-fold pyramid over a quadrilateral . these polytopes have @xmath22 vertices and excess degree @xmath107 . [ rmk : ad - facets]the facets of the @xmath1-polytope @xmath105 , @xmath88 in total , are as follows . * @xmath27 copies of @xmath108 , * 4 simplicial prisms , and * 2 copies of @xmath109 . denote by @xmath110 a polytope obtained by truncating a simple vertex of a @xmath111-triplex . the polytope @xmath112 is the well known _ 5-wedge _ ; it can be obtained as the wedge @xcite at an edge over a pentagon . the polytope @xmath110 can also be realised as follows : the convex hull of @xmath112 and a simplicial @xmath20- prism @xmath48 , where each vertex of one copy of the @xmath113-simplex in @xmath48 is connected to each of the three vertices in a triangle of @xmath112 , and each vertex of the other copy of the @xmath113-simplex in @xmath48 is connected to each of the remaining five vertices of @xmath112 . these polytopes also have @xmath22 vertices and excess degree @xmath107 . [ rmk : bd - facets ] the facets of the @xmath1-polytope @xmath110 , @xmath88 in total , are as follows . * @xmath27 copies of @xmath114 , * 2 simplices , * 1 simplicial prism , * 1 copy of @xmath109 , and * 2 pentasms . denote by @xmath115 a polytope obtained by slicing one _ simple edge _ , i.e. an edge whose endvertices are both simple , of a @xmath106-triplex . it has @xmath116 vertices and excess degree @xmath7 . [ rmk : cd - facets ] the facets of the @xmath1-polytope @xmath115 , @xmath88 in total , are as follows . * @xmath7 copies of @xmath117 , * 3 simplicial prisms , * 1 copy of @xmath118 , and * 1 simplex . for consistency here , we can define @xmath119 as a quadrilateral . denote by @xmath120 a certain polytope which can be expressed as the sum of two simplices . one concrete realisation of it is given by the convex hull of @xmath121 where @xmath122 is the standard basis of @xmath123 . it is easily shown to have @xmath116 vertices ; of these , one has excess degree @xmath7 , and the rest are simple . [ rmk : sigmad - facets ] the facets of the @xmath1-polytope @xmath120 , @xmath88 in total , are as follows . * @xmath8 copies of @xmath124 , * 2 simplicial prisms , * 2 simplices . for consistency , we can also define @xmath125 as a quadrilateral . diagrams of @xmath126 and @xmath127 appear in [ fig:10v-21e](b ) in 6 . let us denote by @xmath128 the result of truncating one vertex from @xmath74 . this is a simple @xmath77-dimensional polytope with @xmath129 vertices and @xmath130 facets . [ rmk : gamma - facets ] the facets of the @xmath131-polytope @xmath128 , @xmath88 in total , are as follows . * @xmath75 copies of @xmath132 , * @xmath76 copies of @xmath133 , * @xmath21 copy of @xmath134 , * @xmath21 copy of @xmath135 , * 1 simplex . denote by @xmath136 the special case @xmath137 , i.e. the polytope obtained by slicing one vertex of a simplicial @xmath1-prism ; it clearly has @xmath138 vertices . observe that @xmath139 is a pentagon , and that @xmath112 coincides with @xmath140 ; so @xmath136 can be considered a generalisation of the 5-wedge . [ rmk : jd - facets ] the facets of the @xmath1-polytope @xmath136 , @xmath88 in total , are as follows . * @xmath8 copies of @xmath141 , * 2 simplicial prisms , * 2 simplices . it can be shown that , for @xmath142 , @xmath136 is the unique simple polytope with @xmath138 vertices ; see [ lem : countsimplep ] below . some of the examples we have just defined coincide in low dimensions . in particular , @xmath143 , @xmath144 and @xmath145 . by definition @xmath146 and @xmath147 . with the fundamental examples now elucidated , we can characterise all the simple @xmath1-polytopes with up to @xmath148 vertices , which will be useful later in the paper . first we need the following result , of independent interest . we omit the case @xmath26 in the statement of the next two results , since all polygons are simple . [ prop : simple - prisms - simplices ] a simple @xmath1-polytope @xmath2 in which every facet is a prism or a simplex is either a @xmath1-simplex , a simplicial @xmath1-prism , @xmath149 or @xmath150 . if @xmath2 is not a simplex , then it is not simplicial , so assume that some facet @xmath29 is a prism . choose a ridge @xmath48 in @xmath29 which is also a prism , and let @xmath42 be the other facet containing @xmath48 . then @xmath42 is also a prism and every vertex in @xmath48 has degree @xmath1 in the subgraph @xmath151 . so no vertex in @xmath48 is adjacent to any vertex outside @xmath151 . if we remove the four vertices in @xmath152 from the graph of @xmath2 , the resulting graph will be disconnected . by balinski s theorem ( * thm . 3.14 ) , we have @xmath25 . denote by @xmath153 the number of triangular 2-faces of @xmath2 , and by @xmath154 the number of quadrilateral 2-faces . if @xmath9 , simplicity implies @xmath155 . then @xmath4 is even and @xmath156 thus either @xmath157 and @xmath2 is a simplex , or @xmath158 and @xmath2 is a prism , or @xmath159 and @xmath2 is a cube . if @xmath160 , simplicity implies @xmath161 , whence @xmath162 . simplicity also ensures that every edge belongs to precisely three 2-faces . by hypothesis , every facet contains at least 2 triangular faces , and every triangular face , being a ridge , belongs to precisely 2 facets . thus @xmath163 . the lower bound theorem @xcite tells us @xmath164 . then @xmath165 thus @xmath166 . so either @xmath167 and @xmath2 is a simplex , or @xmath168 , and [ rmk : dplus2facets ] tells us that @xmath2 has the form @xmath74 for some @xmath85 with @xmath169 . [ lem : countsimplep ] up to combinatorial equivalence , a. the simplex @xmath170 is the only simple @xmath1-polytope with strictly less than @xmath62 vertices ; b. the simplicial prism @xmath171 is the only simple @xmath1-polytope with between @xmath62 and @xmath172 vertices ; c. @xmath173 is the only simple @xmath1-polytope with @xmath174 vertices ; d. the only simple @xmath1-polytope with @xmath116 vertices is the 6-dimensional polytope @xmath175 ; e. the only simple @xmath1-polytopes with @xmath138 vertices are the polytope @xmath136 , the 3-dimensional cube @xmath149 and the 7-dimensional polytope @xmath176 ; f. there is a simple polytope with @xmath148 vertices if and only if @xmath177 or 8 ; the only possible examples are @xmath178 , @xmath179 and @xmath180 . assertions ( i ) to ( v ) are simply a rewording of ( * ? ? ? 10(ii)-(iii ) ) . we prove ( vi ) . recall that two simple polytopes with the same graph must be combinatorially equivalent @xcite . for @xmath160 or 8 , the validity of the three examples given is easy to verify . conversely , suppose @xmath2 is a simple polytope with @xmath148 vertices . then @xmath1 must be even . every facet of @xmath2 is simple , and so has an even number of vertices ( because @xmath8 is odd ) . if there were a facet with @xmath116 vertices , there would be @xmath116 edges running out of it and @xmath181 edges running out of the two external vertices . but @xmath182 . so any facet has at most @xmath183 vertices . we know from ( i ) to ( v ) that any facet must be a simplex , a prism , or have @xmath184 or @xmath172 vertices . first suppose some @xmath29 has @xmath172 vertices . the four vertices outside all have degree @xmath1 , so the number of edges between them and @xmath29 is at least @xmath185 . on the other hand , there are exactly @xmath172 edges running out of @xmath29 . so @xmath186 whence @xmath187 . next suppose some facet @xmath29 has @xmath188 vertices . each of the six vertices outside each has degree @xmath1 , so the number of edges between them and @xmath29 is at least @xmath189 . on the other hand , there are exactly @xmath184 edges running out of @xmath29 . again @xmath187 . now we show that the 3 examples in the first paragraph are the only possibilities when @xmath177 or 8 . if @xmath190 , every facet must be @xmath191 or @xmath176 ; a short calculation shows that @xmath2 must be @xmath180 . if @xmath160 , simplicity ensures that no facet has 10 vertices . thus every facet is either a simplex , a prism , a cube or a 5-wedge . [ prop : simple - prisms - simplices ] ensures that at least one facet is neither a simplex nor a prism . if some facet is a cube , it is not hard to verify that four of the other facets must be prisms , and two of them must be cubes ; this is the only way the 4 vertices outside can be connected to give us the graph of a simple polytope , and the graph is that of @xmath178 . otherwise some facet is a 5-wedge , and there are a number of cases to consider . since each such facet contains two pentagonal faces , and each pentagonal ridge belongs to two such facets , we can find a collection of facets @xmath192 , with each @xmath193 being a 5-wedge , and @xmath194 being a pentagon for each @xmath195 . clearly there must be at least 3 pentagonal faces , so @xmath196 . we can not have @xmath197 , or there would be too many vertices . if @xmath198 , then @xmath199 would contain at least 3 of the vertices in @xmath200 . then the affine hyperplane containing @xmath201 would also contain @xmath200 . this means that all 3 facets lie in the same 3-dimensional affine subspace , which is absurd . so we must have @xmath202 . the graphs of these four facets actually determine the entire graph of @xmath2 ; it is the graph of @xmath179 . finally we need to show that the case @xmath203 does not arise . this was first proved by lee ( * ? ? ? * example 4.4.17 ) , using the @xmath204-theorem , but we give an independent argument . suppose @xmath2 is such a 6-polytope ; then every facet must be either a simplex , a prism , @xmath205 or @xmath206 . consider a ridge @xmath48 which is a 4-prism ; clearly such ridges exist , because @xmath2 is not simplicial . denote by @xmath29 and @xmath42 the two facets containing @xmath48 ; since @xmath207 , the previous result ensures that one of them is not a prism . they can not both be @xmath205 , because then @xmath151 would contain more than 18 vertices . if @xmath29 is @xmath205 and @xmath42 is @xmath206 , then @xmath151 would contain 18 vertices and 49 edges , with the 6 vertices in @xmath49 guaranteeing another 6 edges ; but @xmath208 , contradicting simplicity . if @xmath29 is @xmath205 and @xmath42 is a prism , then @xmath151 would contain 16 vertices and 44 edges , with the 2 vertices outside @xmath151 guaranteeing another 11 edges ; but @xmath209 , again contradicting simplicity . likewise if @xmath29 and @xmath42 are both @xmath206 , then @xmath151 would contain 16 vertices and 44 edges , with the 2 vertices outside @xmath151 guaranteeing another 11 edges . finally if @xmath29 is @xmath206 and @xmath42 is a prism , then @xmath151 would contain 14 vertices and 39 edges , with the 2 vertices outside @xmath151 guaranteeing another 18 edges ; but @xmath210 . this exhausts all the possibilities , so there is no simple 6-polytope with 18 vertices . a polytope @xmath2 is _ ( minkowski ) decomposable _ if it can be written as the minkowski sum of two polytopes , neither of which are homothetic to @xmath2 ; otherwise it is _ indecomposable_. see ( * ? ? ? 15 ) for a more detailed account and historical references . as usual , the _ ( minkowski ) sum _ of two polytopes @xmath211 is defined to be @xmath212 , and a polytope @xmath2 is _ homothetic _ to a polytope @xmath213 if @xmath214 for @xmath215 and @xmath216 . all polygons other than triangles are decomposable ; this topic becomes more serious when @xmath217 . a polytope can have a ( geometric ) realisation that is decomposable and another realisation that is indecomposable . we will generally not distinguish between polytopes which are combinatorially equivalent , i.e. have isomorphic face lattices . smilansky @xcite calls a polytope @xmath2 _ combinatorially decomposable ( resp . indecomposable ) _ if whenever @xmath213 is combinatorially equivalent to @xmath2 , then @xmath213 is also decomposable ( resp . indecomposable ) ; otherwise @xmath2 is called _ conditionally decomposable_. in this paper we only come across polytopes which are combinatorially decomposable , just called decomposable henceforth , or combinatorially indecomposable polytopes , just called indecomposable henceforth . but it is important to be aware that conditionally decomposable polytopes do exist ; see @xcite , ( * ? ? ? 1 ) or ( * ? ? ? * example 11 ) for further discussion . the main tool for establishing decomposability is the following concept . we will say that a facet @xmath29 in a polytope @xmath2 has * shephard s property * if every vertex in @xmath29 has a unique neighbour outside @xmath29 . this is not an intrinsic property of @xmath29 , but rather of the way that @xmath29 sits inside @xmath2 . this concept appears implicitly in the proof of ( * ? ? ? * ( 15 ) ) , and leads to the following result . [ thm : shp ] a polytope @xmath2 is decomposable whenever there is a facet @xmath29 with shephard s property , and @xmath2 has at least two vertices outside @xmath29 . in other words , if a polytope @xmath2 has a facet @xmath29 with shephard s property , then it is either decomposable or a pyramid with base @xmath29 . this result was essentially proved by shephard ( * ? ? ? * result ( 15 ) ) . he made the stronger assumption that every vertex in @xmath29 is simple in @xmath2 , but the general statement does follow from his proof . another proof appears in ( * ? ? ? we will say that a polytope is a _ shephard polytope _ if it has at least one facet with shephard s property . it is not hard to see that a polytope is simple if and only if every facet has shephard s property . [ thm : shp ] implies that any simple polytope is either decomposable or a simplex . later we will generalise this , to show that any polytope with small excess is either decomposable , or a pyramid . the following related concept will be useful in 4 . let us say that a facet @xmath29 in a polytope @xmath2 has * kirkman s property * if every other facet @xmath42 intersects @xmath29 at a ridge . again this is not an intrinsic property of @xmath29 alone . lee @xcite calls a _ kirkman polytope _ any polytope in which at least one facet has kirkman s property ; klee @xcite had earlier made an equivalent definition for simple polytopes . the generalisation to non - simple polytopes in @xcite is quite natural . following klee , we call a _ super - kirkman polytope _ any polytope in which _ every _ facet has kirkman s property . this is equivalent to the dual polytope being 2-neighborly . numerous examples ( e.g. the cube ) show that shephard s property does not imply kirkman s property . in the other direction , if @xmath2 is a pyramid over @xmath74 , where @xmath218 , then @xmath2 is a super - kirkman polytope , but only one facet has shephard s property . as a common weakening of both properties , we could say that a facet @xmath29 in a polytope @xmath2 has the weak kirkman - shephard property if every other facet @xmath42 is either disjoint from @xmath29 , or intersects @xmath29 at a ridge . although this defines a natural property , we have no applications for it so far . obviously a polytope is semisimple if every facet has the weak kirkman - shephard property . turning now to indecomposability , a much bigger toolkit of sufficient conditions is available . rather than giving a detailed survey of this topic , we will just summarize the results we need . the main concept we need is due to kallay @xcite ; he defined a _ geometric graph _ as any graph @xmath42 whose vertex set @xmath219 is a subset of a finite - dimensional real vector space @xmath220 , and whose edge set @xmath221 is a subset of the line segments joining members of @xmath219 . then a function @xmath222 is called a _ decomposing function _ for the graph @xmath223 if it has the property that @xmath224 is a scalar multiple of @xmath225 for each edge @xmath226\in e$ ] . a geometric graph @xmath227 is then called _ decomposable _ if there is a decomposing function which is not the restriction of a homothety on @xmath220 . if the only decomposing functions are homotheties then @xmath42 is called indecomposable . kallay showed ( * ? ? ? 1a ) that a polytope is decomposable if and only if its edge graph is decomposable in this sense . the study of geometric graphs is actually a useful tool for establishing indecomposability . for a detailed proof of the next result , see @xcite and the references therein . [ thm : mcmullen - decomp ] a. if the vertices of a geometric graph are affinely independent , and its edges form a cycle , then the graph is indecomposable . b. if @xmath228 and @xmath229 are two indecomposable geometric graphs with ( at least ) two vertices in common , then the graph @xmath230 is also indecomposable . c. if @xmath2 is a polytope , @xmath42 is a subgraph of the skeleton of @xmath2 , @xmath42 contains at least one vertex from every facet of @xmath2 , and @xmath42 is an indecomposable graph , then @xmath2 is an indecomposable polytope . d. conversely , if @xmath2 is an indecomposable polytope , then its skeleton @xmath42 is an indecomposable geometric graph . a hamiltonian cycle through the vertices of a simplex is an affinely independent cycle ; this is one way to see that a simplex is indecomposable . repeated application of part ( ii ) then shows that any simplicial polytope is indecomposable . in particular , the dual of any simple polytope must be indecomposable . the corollary below generalises that . it depends on the following result , which seems to be a basic and fundamental result on decomposability , but we could not find a reference . [ thm : d-1decomposablefacets ] let @xmath2 be a @xmath1-polytope with strictly less than @xmath1 decomposable facets . then @xmath2 is indecomposable . let @xmath76 denote the number of decomposable facets of @xmath2 , with @xmath231 . consider the corresponding collection of vertices in the dual polytope @xmath232 . thanks to balinski s theorem ( * ? ? ? * thm . 3.14 ) , their removal does not disconnect @xmath232 , so all the other vertices ( which correspond to indecomposable facets of @xmath2 ) can be ordered into a sequence , possibly with repetition , so that any successive pair defines an edge of @xmath232 . this means that the indecomposable facets of @xmath2 form a strongly connected family , in the sense that each successive pair intersects in a ridge . this is much more than we need ; intersecting in an edge gives us a suitably connected family of indecomposable faces . the union of this family is clearly an indecomposable graph . it touches every facet because there are only @xmath76 other facets , and any facet intersects at least @xmath1 other facets . hence [ thm : mcmullen - decomp ] gives the conclusion . furthermore , this is best possible : there are decomposable @xmath1-polytopes with precisely @xmath1 decomposable facets , namely @xmath1-prisms and capped @xmath1-prisms . [ cor : dualsexcessd-2-d-1-indecomp ] let @xmath2 be @xmath1-polytope at most @xmath8 nonsimple vertices . then the dual polytope is indecomposable . [ thm : d-1decomposablefacets ] gives the result at once , as duals of @xmath1-polytopes with at most @xmath8 nonsimple vertices have at most @xmath8 decomposable facets . this is also best possible . the bipyramid over a @xmath34-simplex has exactly @xmath1 nonsimple vertices , but its dual is the prism , which is decomposable . following grnbaum s notation @xcite , define @xmath233 it is convenient to note that for @xmath68 @xmath234 [ thm : triplexes ] let @xmath2 be a @xmath1-polytope with @xmath70 vertices with @xmath68 . then @xmath2 has at least @xmath235 edges . equivalently , we can say that @xmath236 . furthermore , equality is obtained only if @xmath2 is a @xmath65-triplex . [ thm : pentasm ] the polytopes with @xmath12 vertices and @xmath100 or fewer edges are as follows . a. for @xmath9 , there are exactly two polyhedra with 7 vertices and 11 edges ; the pentasm , and @xmath237 . none have fewer edges . b. for @xmath160 , a sum of two triangles @xmath150 is the unique polytope with 18 edges , and the pentasm is the unique polytope with 19 edges . none have fewer edges . c. for @xmath238 , the pentasm is the unique polytope with @xmath100 edges . none have fewer edges . we will denote by @xmath239 the set of possible values of the excess of all @xmath1-dimensional polytopes . the main result here is that the smallest two values in @xmath239 are 0 and @xmath7 ; nothing in between is possible . if @xmath240 , then @xmath241 , so [ lem : non - disjointfacets ] establishes this for any polytope which is not semisimple . accordingly , we restrict our attention to semisimple polytopes . first consider the case in which every facet is simple . [ prop : charpsimple ] let @xmath2 be a semisimple @xmath1-polytope in which every facet is simple . then @xmath2 is simple . this is immediate from [ rmk : simplenonsimplevertex ] . in addition to the assumption that every two nondisjoint facets intersect at a ridge , thanks to [ prop : charpsimple ] , we can now assume that our nonsimple @xmath1- polytope @xmath2 contains a nonsimple facet . [ lem : facet - not - all - excess ] the excess of a nonsimple polytope is strictly larger than the excess of any of its facets . suppose otherwise . let @xmath29 be a facet with equal excess to the polytope . then , every vertex outside @xmath29 is simple . take a nonsimple vertex @xmath16 in @xmath29 and neighbour @xmath38 of @xmath16 lying in @xmath39 . since @xmath38 is simple the edge @xmath41 is contained in exactly @xmath8 facets , but @xmath16 must be contained in at least @xmath40 facets . thus , @xmath16 has at least two neighbours outside @xmath29 ( cf . [ rmk : simplevertexoutside ] ) , implying the excess of the polytope is larger than the excess of @xmath29 . we are now ready to prove the theorem . [ thm : excess ] let @xmath2 be a @xmath1-polytope . then the smallest values in @xmath239 are 0 and @xmath7 . proceed by induction on @xmath1 , with the base case @xmath9 being easy . a tetrahedron is simple , and square pyramid has excess 1 . if @xmath2 is a simple polytope we get excess 0 , so assume that @xmath2 is nonsimple . by virtue of [ lem : non - disjointfacets ] we can assume that every nondisjoint pair of facets intersect at a ridge , and by [ prop : charpsimple ] that there exists a nonsimple facet @xmath29 with excess at least @xmath27 by the inductive hypothesis . if there is a nonsimple vertex outside @xmath29 , we are done . if a nonsimple vertex of @xmath2 is simple in @xmath29 , we are also at home . so we can further assume that the facet @xmath29 contains all the nonsimple vertices of @xmath2 and that each nonsimple vertex @xmath16 of @xmath2 is nonsimple in @xmath29 and has exactly one neighbour @xmath38 outside @xmath29 . in this case , the facet @xmath29 would contain all the excess of the polytope , which is ruled out by [ lem : facet - not - all - excess ] . the existence , in every dimension , of @xmath1-polytopes with excess @xmath7 and @xmath1 has been observed in 2 . the existence of @xmath1-polytopes with excess @xmath64 , e.g. the cyclic polytope with @xmath64 vertices , is also well known . ( the characterisation of @xmath1-polytopes with @xmath64 vertices @xcite gives many more examples . ) in the next section , we will show that polytopes with excess @xmath8 only exist in dimensions 3 and 5 . what about higher values ? it is clear that the excess degree can be any natural number if @xmath9 , and any even number if @xmath160 . we can show that the excess degree takes all possible values above @xmath7 if @xmath242 or @xmath243 . we suspect that there are gaps in the possible values of @xmath0 for @xmath244 . for @xmath245 , we can show that the excess degree takes all possible values from 7 onwards , expect perhaps 11 . apart from the excess @xmath8 , we have so far been unable to prove the existence of any other such gap above @xmath7 . the next theorem shows that for each dimension , the number of further gaps , if any , is finite . [ lem : excess - few - gaps - facets ] if the @xmath1-polytope @xmath2 is a pyramid with base @xmath29 , and @xmath29 has @xmath37 vertices , then @xmath246 . [ thm : excessfewgaps ] if @xmath1 is even , then @xmath239 contains every even integer in the interval @xmath247 . if @xmath1 is odd , then @xmath239 contains every integer in the interval @xmath248 . according to @xcite , for any integer @xmath249 and any integer @xmath36 in the interval @xmath250 $ ] there is a 4-polytope with @xmath37 vertices and @xmath36 edges . ( a stronger assertion can be made if we exclude the values @xmath251 , but we do not need it here . ) it follows that for any @xmath249 and any even @xmath252 in the interval @xmath253 $ ] there is a 4-polytope with @xmath37 vertices and excess @xmath252 . let us denote by @xmath254 the collection of all integers in the interval @xmath255\ ] ] which have the * same parity * as the two endpoints . applying [ lem : excess - few - gaps - facets ] to @xmath113-fold pyramids , we see that for any @xmath207 , @xmath256 and any @xmath257 , there is a @xmath1-polytope with @xmath258 vertices and excess @xmath252 . we now consider three cases , depending on the value of @xmath1 . first consider the case when @xmath238 is odd . let @xmath90 be the smallest integer greater than @xmath259 ; clearly @xmath260 . for any @xmath261 it is clear that @xmath239 contains the ( parity based ) interval @xmath254 . it is easily seen that @xmath262 , so @xmath263 , whence @xmath264 , i.e. the intervals @xmath254 and @xmath265 , which have elements of the same parity , overlap . consider separately the intervals @xmath266 , whose elements all have the same parity , and the intervals @xmath267 whose elements all have the other parity . since @xmath268 , a moment s reflection shows that @xmath239 contains every integer , both even and odd , from @xmath269 onwards . but @xmath270 , and @xmath271 , so @xmath272 thus @xmath239 contains every integer greater than @xmath273 now consider the case when @xmath274 is even . let @xmath90 be the smallest integer greater than or equal to @xmath275 . then @xmath276 . for any @xmath261 we have @xmath277 it follows that @xmath278 , which means that @xmath279 , i.e. there is no gap between the parity - based intervals @xmath254 and @xmath280 . each such interval is contained in @xmath239 , which thus contains all even integers in @xmath281 . on the other hand , @xmath282 , and @xmath283 , so @xmath284 thus @xmath239 contains every even integer greater than or equal to @xmath285 . for lower dimensions , we recall that if @xmath89 and @xmath252 is an even number in the interval @xmath286 $ ] , then there is a @xmath1-polytope @xmath2 with excess @xmath252 ( and @xmath88 vertices ) . this is a reformulation of ( * ? ? ? * lemma 14 ) ; a tiny modification of the proof there shows that @xmath2 can be chosen to have at least one ( in fact every ) facet being a simplex . now stacking a vertex on a simplex facet will increase the excess by precisely @xmath1 . thus we obtain @xmath1-polytopes whose excesses are every even number in the intervals @xmath287 $ ] , @xmath288 $ ] , @xmath289 $ ] etc . if @xmath290 , the union of these intervals in @xmath291 ; clearly @xmath292 , so this completes the proof . we now note that the excess degree takes all possible values above @xmath7 if @xmath1 is sufficiently small . in particular , if @xmath293 , the only impossible values of @xmath239 are those excluded by [ thm : excess ] and parity considerations . in case @xmath294 or 10 , the proof just given shows that every even integer from @xmath107 onwards is the excess degree of some polytope ; and the existence of @xmath1-polytopes with excess @xmath295 and @xmath64 has already been noted . it is clear then that every even value above @xmath7 is realised as the excess of some polytope . in case @xmath242 , this proof shows that every integer from 7 onwards is the excess degree of some polytope ; the existence of 5-polytopes with excess 3 or 5 should be clear . the triplex @xmath296 has excess degree 4 , and a 3-fold pyramid over a pentagon has excess degree 6 . [ cor : dualsexcessd-2-d-1-indecomp ] showed the indecomposability of the duals of @xmath1-polytopes with at most @xmath8 nonsimple vertices . in this section , we establish stronger conclusions for polytopes with small excess , namely @xmath297 or @xmath8 . we show that this imposes strong restrictions on the distribution of nonsimple vertices . a @xmath1-polytope with excess @xmath7 either has a single vertex with excess @xmath7 , or @xmath7 vertices each with excess 1 ; in the latter case , the nonsimple vertices form a @xmath20-face . a polytope can have excess @xmath8 only if @xmath9 or 5 , and can not have 3 nonsimple vertices . thus there is at least one further gap in the possible values of the excess degree , from dimension 7 onwards . we begin by examining the intersection patterns of facets ; these are also severely restricted by the assumption of low excess . with the exception of ten 3-dimensional polyhedra , every @xmath1-polytope with excess @xmath7 or @xmath8 is a shephard polytope . nine of these ten turn out to be ( minkowski ) decomposable . thus every polytope with low excess is either decomposable or a pyramid or @xmath61 ( see [ fig : polytopes](d ) ) ; and we will see that in all cases their duals are indecomposable . of course a polytope with excess 0 is either decomposable or a simplex . thus the excess theorem is highly applicable to questions about the decomposability of polytopes . the results just mentioned are best possible since there are @xmath1-polytopes with excess @xmath1 which are decomposable and @xmath1-polytopes that are indecomposable . for instance , the capped prism @xmath298 is decomposable and self - dual , a bipyramid over a @xmath34-simplex is indecomposable with decomposable dual , and the capped prisms @xmath94 for @xmath299 are decomposable with indecomposable duals . for @xmath9 , all polyhedra with 7 vertices and 7 faces have excess @xmath1 ; with the exception of the capped prism , they are all indecomposable with indecomposable duals . for simplicity of expression , we will often simply state the conclusion that we have a shephard polytope ; bear in mind that [ thm : shp ] then guarantees that the polytope is either decomposable or a pyramid . we begin by making explicit the following easy consequence of [ lem : non - disjointfacets ] . [ lem : facetintersection ] let @xmath29 and @xmath42 be distinct nondisjoint facets of a polytope @xmath2 and set @xmath300 . if @xmath301 , then @xmath51 is either 0 , @xmath27 or @xmath7 . if @xmath302 , then either @xmath303 or @xmath7 ; or @xmath304 and @xmath242 . in case @xmath305 , @xmath44 is either a simplex or a quadrilateral . let @xmath67 be the number of vertices in @xmath44 ; then [ lem : non - disjointfacets ] informs us that @xmath306 . clearly @xmath307 is one possibility . henceforth , it is enough consider only the case @xmath240 . the inequality @xmath308 is equivalent to @xmath309 . the only solutions for this are @xmath310 , @xmath311 and @xmath312 . clearly @xmath44 is a simplex if @xmath310 or 1 . if @xmath311 , our original inequality becomes @xmath313 . if @xmath301 , then @xmath314 and again @xmath44 is a simplex . in case @xmath302 , it is also possible that @xmath315 . but then the contribution to the excess of the edges leaving @xmath44 is already @xmath8 , meaning that @xmath44 must be simple . this is only possible if @xmath316 , and then @xmath202 and @xmath242 . this result enables us to study the structure of polytopes with excess degree @xmath7 . for 3-dimensional polyhedra this is quite simple . any such polyhedron has a unique nonsimple vertex , which has degree 4 , and so the polyhedron is either a simplex or decomposable . we now proceed to higher dimensions . [ lem : excessd-2facets0 ] let @xmath2 be @xmath1-polytope with excess degree @xmath7 . if @xmath2 has a unique nonsimple vertex ( necessarily with excess @xmath7 ) , then it has two nondisjoint facets intersecting at just this vertex . conversely , if @xmath317 and @xmath318 be any two nondisjoint facets of @xmath2 such that @xmath319 , then this vertex is the unique nonsimple vertex ( with excess @xmath7 ) , and consequently @xmath2 is a shephard polytope . furthermore , @xmath2 has a facet with excess @xmath27 . suppose there is a unique nonsimple vertex @xmath16 with degree @xmath181 . thanks to [ lem : vertex - all - neighbours - simple ] , the vertex figure @xmath320 of @xmath2 at @xmath16 is a simple @xmath34-polytope , in fact a simplicial prism . let @xmath321 and @xmath322 be the two opposite simplices in @xmath320 . then the two facets containing @xmath16 arising from @xmath321 and @xmath322 intersect only at @xmath16 . conversely , suppose @xmath323 . then @xmath16 has degree @xmath8 in both facets , and excess @xmath7 . thus every other vertex in @xmath2 is simple . there is a facet in @xmath2 which does not contain @xmath16 , which clearly has shephard s property . finally , let @xmath29 be the other facet arising from a ridge of @xmath317 which contains @xmath16 . the facet @xmath29 must then intersect @xmath318 at a ridge ; it ca nt intersect @xmath318 at an edge or a @xmath20-face since @xmath16 is the only nonsimple vertex ( cf . [ lem : non - disjointfacets ] ) . the degree of @xmath16 in @xmath29 is @xmath324 . [ lem : excessd-2facetsd-3 ] let @xmath2 be @xmath1-polytope with excess degree @xmath7 . let @xmath317 and @xmath318 be any two facets of @xmath2 such that @xmath325 has dimension @xmath27 . then a. @xmath58 is a simplex , and each of its @xmath7 vertices has excess degree one in @xmath2 . b. @xmath2 is a shephard polytope . hence it is either decomposable , or a @xmath326-fold pyramid . c. every facet in @xmath2 intersecting @xmath58 but not containing it misses exactly one vertex of @xmath58 and every vertex of @xmath58 in the facet has degree @xmath1 ; thus the facet has excess @xmath27 . d. there are precisely four facets containing @xmath58 , and each of them is simple . the result is clearly true for @xmath9 , so assume @xmath89 . from [ lem : non - disjointfacets ] it follows that every vertex in @xmath58 has degree @xmath40 in @xmath2 , that @xmath58 is a simplex , and that every other vertex in @xmath2 is simple . moreover every vertex outside @xmath58 is simple in @xmath2 , while every vertex in @xmath58 is simple in both @xmath317 and @xmath318 , and has no neighbours outside @xmath53 . since @xmath327 , we can find distinct vertices @xmath328 . consider a facet @xmath29 containing @xmath16 but not @xmath37 ; we claim that any such facet has shephard s property . note that @xmath29 intersects @xmath317 and @xmath318 at a ridge , and contains every vertex in @xmath58 but @xmath37 , every neighbour of @xmath16 in @xmath329 , and every neighbour of @xmath16 in @xmath330 . the same argument applies to every vertex in @xmath331 . this in turn implies that every vertex in @xmath29 has exactly one neighbour outside @xmath29 . in other words , the @xmath27 vertices in @xmath331 has degree @xmath1 in @xmath29 , and hence the facet @xmath29 has excess @xmath27 . thus , by [ thm : shp ] @xmath2 is decomposable unless @xmath332 , in which case @xmath2 is a pyramid over @xmath29 with apex @xmath38 . as there is nothing special about @xmath16 or @xmath37 , in the case of @xmath2 being indecomposable , we see that @xmath2 is a pyramid with every vertex in @xmath58 acting as an apex , that is , @xmath2 is a @xmath326-fold pyramid . this completes the proof of ( iii ) . for ( iv ) , fix a vertex @xmath333 , let @xmath334 and @xmath335 be the neighbours of @xmath37 in @xmath329 , and let @xmath336 and @xmath337 be the neighbours of @xmath37 in @xmath330 . for each @xmath195 , denote by @xmath338 the smallest face containing @xmath339 . since @xmath58 has codimension two in @xmath317 , it must be the intersection of 2 ridges in @xmath317 ( i.e. facets of @xmath317 ) . thus @xmath321 and @xmath322 must be ridges of @xmath2 . likewise @xmath340 and @xmath341 must be ridges of @xmath2 . these four ridges are distinct , so there must be at least four facets of @xmath2 which contain @xmath58 . we need to show that there are only four , not five or six . denote by @xmath342 the smallest face of @xmath2 containing @xmath343 for each @xmath344 . obviously any facet containing @xmath58 must be of this form , and @xmath345 and @xmath346 . with respect to the inclusion @xmath347 , the other facet containing @xmath321 must be either @xmath348 or @xmath349 ; without loss of generality suppose it is @xmath348 . since @xmath350 also , the unique representation of ridges as the intersection of two facets implies that @xmath349 can not be a facet . continuing , we conclude that @xmath351 and that @xmath352 is the fourth facet containing @xmath58 . since @xmath353 , both these facets are also simple . we can now illuminate the structure of polytopes with excess @xmath7 . [ lem : excess - d-2 - 4dim ] let @xmath2 be a 4-polytope with excess degree two . then @xmath2 is not a super - kirkman polytope . if there is a unique nonsimple vertex , this follows from [ lem : excessd-2facets0 ] . otherwise , @xmath2 has two vertices , say @xmath16 and @xmath37 , of degree five . they must be connected by an edge ; otherwise , considering a facet @xmath42 containing @xmath16 , but not @xmath37 , would give that @xmath16 is nonsimple in @xmath42 ( by [ rmk : simplenonsimplevertex ] ) with at least two neighbours outside the facet ( by [ rmk : simplevertexoutside ] ) . consider a facet @xmath29 containing the edge @xmath354 . then both @xmath16 and @xmath37 must be nonsimple in @xmath29 by [ rmk : simplenonsimplevertex ] , which implies that the excess of @xmath29 is two , contradicting [ lem : facet - not - all - excess ] . in the next result , examples for case ( i ) include @xmath355 , @xmath120 and @xmath356 . examples for case ( ii ) include @xmath357 , @xmath115 , @xmath87 and pentasms . [ thm : excess - d-2-full - story ] any @xmath1-polytope @xmath2 with excess exactly @xmath7 either a. has a unique nonsimple vertex , which is the intersection of two facets , or b. has @xmath7 vertices of excess degree one , which form a @xmath20-simplex which is the intersection of two facets . in either case , @xmath2 is a shephard polytope and has a facet with excess @xmath27 . we proceed by induction on the dimension . the case @xmath9 is obvious , with ( i ) and ( ii ) coinciding . the case @xmath160 is settled by [ lem : facetintersection ] and [ lem : excess - d-2 - 4dim ] . consequently , assume henceforth that @xmath238 . [ lem : excessd-2facets0 ] and [ lem : excessd-2facetsd-3 ] establish the conclusion if there are two facets which intersect in a vertex or a @xmath20-face . so we may assume that @xmath2 is a super - kirkman polytope ; we will show that this case actually does not arise . by [ prop : charpsimple , thm : excess ] , we can assume that there exists a facet @xmath29 with excess at least @xmath27 , and by [ lem : facet - not - all - excess ] that @xmath29 ( and every nonsimple facet ) has excess exactly @xmath27 . by induction , we may suppose that @xmath29 satisfies either ( i ) or ( ii ) . first suppose that the facet @xmath29 contains a nonsimple vertex @xmath37 , with degree @xmath358 therein , and two ridges @xmath48 and @xmath359 such that @xmath360 . let @xmath42 be the other facet corresponding to @xmath48 . then @xmath37 is not simple in @xmath42 because it is not simple in @xmath2 ( cf . [ rmk : facets - intersecting - at - ridges ] ) . consequently , @xmath37 must have degree @xmath1 in @xmath42 , degree @xmath361 in @xmath2 and is the only nonsimple vertex of @xmath2 . this in turn implies that the excess of @xmath42 is @xmath362 , a contradiction . next suppose that @xmath29 contains two ridges @xmath48 and @xmath359 such that @xmath363 is a @xmath113-simplex , with every vertex having excess one in @xmath29 . if there were no other nonsimple vertex in @xmath2 , then every vertex in @xmath58 would be adjacent to some simple vertex in @xmath39 , and hence to at least two vertices outside @xmath29 ( by [ rmk : simplevertexoutside ] ) , meaning the excess degree of @xmath2 would be at least @xmath364 . so there must another nonsimple vertex @xmath16 in @xmath39 , with excess one , and it will be the unique neighbour outside @xmath29 of every vertex in @xmath58 . this implies that the other facet @xmath42 corresponding to @xmath48 must contain @xmath16 . but then @xmath42 contains @xmath7 nonsimple vertices , each having excess degree one in @xmath42 , which is impossible by [ lem : facet - not - all - excess ] . we now turn our attention to the decomposability of @xmath1-polytopes with excess @xmath8 . [ rmk : nonpyramidalindecomp3-polytopes ] since 3-dimensional polyhedra with excess degree @xmath8 do not behave as neatly as those in higher dimensions , we will examine this case in detail first . catalogues of 3-dimensional polyhedra help us to clarify the situation . numbers in ( iv ) in the next statement refer to the catalogue of federico @xcite . the 3-polytope @xmath61 has excess @xmath365 , yet is neither a pyramid nor decomposable ; we will see shortly that it is the only such example . it has two non - simple vertices , and every facet contains at least one of them , so it also fails shephard s property . [ lem : dim3excess2 ] let @xmath2 be @xmath366-polytope whose excess degree is @xmath367 . then either a. @xmath2 is a pentagonal pyramid , which is indecomposable , or b. @xmath2 is @xmath61 , which is also indecomposable but not a pyramid , or c. @xmath2 has shephard s property but is not a pyramid , and hence is decomposable , or d. @xmath2 is one of @xmath368 , @xmath369 , @xmath370 , @xmath371 , @xmath372 , @xmath373 , @xmath374 , @xmath375 , @xmath376 , which are all decomposable , but fail shephard s property . excess two means that for some @xmath67 , @xmath2 has @xmath377 vertices and @xmath378 edges . either one vertex has degree five , or two vertices have degree four . in either case , the number of faces containing a non - simple vertex is at most eight . so if @xmath379 , then @xmath2 has @xmath380 facets , one of which contains only simple vertices , and so has shephard s property . if a pyramid has excess @xmath367 , its base must be a pentagon . if @xmath381 or 4 , then @xmath382 , and @xmath2 must be decomposable by ( * ? ? ? * theorem 6.11(a ) ) . the examples which fail shephard s property can be verified from catalogues ; they are as listed . if @xmath198 , then @xmath2 is @xmath61 or a pentagonal pyramid . we have seen that for a polytope with excess @xmath7 , the non - simple vertices form a face . we will see that this is also true for polytopes with excess @xmath8 , provided @xmath60 . for @xmath9 , it is false ; in every example in case ( iv ) above , the two nonsimple vertices are not adjacent . truncating simple vertices then yields many more examples which fall into case ( iii ) . [ rmk : excessd-1 ] if a @xmath1-polytope has excess @xmath8 , then @xmath1 must be odd . if also @xmath383 then any two nondisjoint facets intersect , like in the case of excess @xmath7 , either at a vertex , a @xmath20-face or a ridge ( as a consequence of [ lem : non - disjointfacets ] ) . if @xmath242 intersections at an edge can also occur . [ lem : excessd-1facets0 ] let @xmath2 be @xmath1-polytope with excess degree @xmath8 , and two facets @xmath317 and @xmath318 intersecting at a vertex . then @xmath9 , and the polytope is as described in [ lem : dim3excess2 ] . let @xmath323 . we consider two cases : either @xmath16 is the only nonsimple vertex in @xmath2 or there is another nonsimple vertex @xmath37 in @xmath2 . in the former case , every vertex in @xmath2 other than @xmath16 is simple and the vertex figure of @xmath2 at @xmath16 is a simple @xmath34-polytope with @xmath384 vertices ( cf . [ lem : vertex - all - neighbours - simple ] ) , so [ lem : countsimplep](ii ) forces @xmath9 or 5 . assume @xmath242 . the vertex figure @xmath320 has @xmath385 vertices , and by [ lem : countsimplep](iii ) can only be @xmath150 ( cf . [ fig : polytopes](c ) ) , which has the property that any two 3-faces intersect at a 2-face . thus , every two facets in @xmath2 containing @xmath16 must intersect at a ridge , a contradiction . in the latter case , the vertex @xmath16 has degree @xmath386 and @xmath37 has degree exactly @xmath40 . [ lem : vertex - all - neighbours - simple ] ensures that @xmath16 and @xmath37 are adjacent . without loss of generality , assume @xmath387 . let @xmath29 be any other facet in @xmath2 containing @xmath37 . if the intersection of @xmath29 and @xmath317 is not a ridge , then [ lem : non - disjointfacets ] tell us that every vertex in @xmath388 is nonsimple , meaning @xmath37 is the only element therein , with excess @xmath7 . this is impossible when @xmath60 . so every facet containing @xmath37 must intersect @xmath317 at a ridge . [ rmk : simplenonsimplevertex ] in turn implies that @xmath37 is nonsimple in @xmath317 . if we choose @xmath29 to contain @xmath37 but not @xmath16 , then [ thm : excess ] gives @xmath389 , again forcing @xmath9 . thus , @xmath9 in both cases . finally note that if a 3-polytope is nonsimple , then a nonsimple vertex must lie in the intersection of a pair of facets , so all such examples are described in [ lem : dim3excess2 ] . we now deal with the particular situation which is mentioned in [ rmk : excessd-1 ] . [ lem : excess4facets1 ] let @xmath2 be @xmath390-polytope with excess degree @xmath391 . let @xmath317 and @xmath318 be any two facets of @xmath2 such that @xmath392 . then @xmath2 is a shephard polytope . let @xmath393 $ ] . then @xmath16 and @xmath37 are both simple in both @xmath317 and @xmath318 , and every other vertex in the polytope is simple . consider a facet @xmath29 containing @xmath16 but not @xmath37 . then , in view of [ lem : facetintersection ] and [ lem : excessd-1facets0 ] we can assume that @xmath29 intersects each @xmath317 and @xmath318 at a ridge . this implies that @xmath16 has exactly one neighbour outside @xmath29 . since every vertex in @xmath29 is simple , @xmath29 has shephard s property , as required . the next two lemmas deal with the case of two nondisjoint facets intersecting at a @xmath20-face or only at ridges . [ lem : excessd-1facetsd-3 ] let @xmath2 be @xmath1-polytope with excess degree of @xmath8 , with @xmath60 . let @xmath317 and @xmath318 be any two facets of @xmath2 whose intersection is a subridge . then @xmath242 , the face @xmath394 is a quadrilateral and @xmath2 is a shephard polytope . recall that @xmath1 must be odd . let @xmath58 denote the subridge @xmath395 . we consider two cases : either ( 1 ) the excess degree of @xmath2 comes only from the nonsimple vertices in @xmath58 , or ( 2 ) the excess degree of @xmath2 comes from the nonsimple vertices in @xmath58 and a nonsimple vertex @xmath37 outside @xmath58 . note that , in the first case , @xmath58 must have @xmath7 vertices , unless @xmath242 , in which case it could have @xmath8 vertices . indeed , suppose @xmath58 has @xmath8 vertices . from [ lem : non - disjointfacets ] it follows that every vertex in @xmath58 has degree @xmath40 in @xmath2 and that @xmath58 is a simple polytope , but the last assertion is only possible if @xmath58 is 2-dimensional , i.e. @xmath396 . so in this particular case @xmath24 and @xmath58 is a quadrilateral . * case 1 . * the excess degree of @xmath2 comes only from the ( nonsimple ) vertices in @xmath58 . we will first rule out this case when @xmath58 has @xmath7 vertices . then it is a simplex , and there exists a vertex in @xmath58 , say @xmath16 , with degree @xmath64 in @xmath2 , while every other vertex in @xmath58 has degree @xmath40 and is simple in both @xmath317 and @xmath318 . if @xmath16 were nonsimple in @xmath317 or @xmath318 , then @xmath397 or @xmath398 by the excess theorem . so @xmath16 is simple in @xmath317 and @xmath318 . denote by @xmath37 the neighbour of @xmath16 outside @xmath53 . we consider the set @xmath399 of facets containing @xmath16 . since @xmath16 is simple in @xmath317 , there are exactly @xmath8 facets in @xmath399 intersecting @xmath317 at a ridge . out of these @xmath8 facets , there are @xmath27 facets which miss one vertex in @xmath58 ; consider one such facet @xmath29 . the facet @xmath29 must also intersect @xmath318 at a ridge , otherwise the excess would be much larger . consequently , every vertex in @xmath400 ( of which there are @xmath401 ) has excess 1 in @xmath29 , i.e. has @xmath1 neighbours in @xmath29 . if @xmath37 belonged to @xmath29 , then @xmath16 would have @xmath40 neighbours in @xmath29 and @xmath402 . however , since @xmath8 is even , @xmath403 can not be odd . thus , @xmath404 . suppose there is a facet @xmath42 in @xmath399 which intersects @xmath317 at @xmath58 ; by [ lem : excessd-1facets0 ] the dimension of the intersection can not be smaller than @xmath27 . now consider any vertex in @xmath58 other than @xmath16 ; its two neighbours in @xmath329 can not be in @xmath42 , so @xmath42 contains both neighbours in @xmath318 . this means that @xmath42 must be @xmath318 . consequently , there are exactly @xmath40 facets in @xmath399 : @xmath317 , @xmath8 facets intersecting @xmath317 at a ridge , and @xmath318 . hence there are at most @xmath367 facets in @xmath399 containing the vertex @xmath37 . but there must be exactly @xmath8 facets in @xmath399 containing @xmath37 . so @xmath405 . for the particular case of @xmath242 and @xmath58 having @xmath8 vertices , every vertex in @xmath58 has excess degree 1 , so we can choose @xmath16 to be any of them . let @xmath406 be a vertex in @xmath58 ( this step requires @xmath60 ) . consider a facet @xmath29 containing @xmath93 but not @xmath16 . then from [ rmk : excessd-1 ] it ensues that @xmath29 intersects each @xmath317 and @xmath318 at a ridge . thus @xmath29 contains every neighbour of @xmath93 in @xmath329 and @xmath330 . the same argument applies to every vertex @xmath407 in @xmath331 . this in turn implies that @xmath29 has shephard s property . we next rule out the second case for @xmath60 . * case 2 . * every vertex in @xmath58 has degree @xmath40 and is simple in both @xmath317 and @xmath318 , and there is a nonsimple vertex @xmath408 of degree @xmath40 . if the vertex @xmath37 does not belong to @xmath53 then all the neighbours of @xmath37 are simple in @xmath2 . from [ lem : vertex - all - neighbours - simple ] it ensues that the vertex figure of @xmath2 at @xmath37 is a simple @xmath34-polytope with @xmath409 vertices , which implies @xmath26 by [ lem : countsimplep ] . so @xmath37 belongs to either @xmath329 or @xmath330 , say @xmath329 , and must be adjacent to at least one vertex @xmath410 . we claim that every vertex @xmath406 in @xmath58 ( which do exist since @xmath60 ) must be adjacent to @xmath37 . suppose otherwise . consider a facet @xmath29 containing @xmath37 but not @xmath16 ( cf . [ rmk : two - vertices ] ) . then @xmath29 must intersect @xmath317 at a ridge ( note that @xmath411 ) . since @xmath317 can not have excess degree precisely 1 , @xmath37 must be simple in @xmath317 . thus there is only one such facet @xmath29 , thereby implying that there are at least @xmath1 facets containing the edge @xmath354 . the edge @xmath354 is contained in exactly @xmath7 facets of @xmath317 ( which are of courses ridges in @xmath2 ) . thus there are at least two facets @xmath228 and @xmath229 containing the edge @xmath354 intersecting @xmath317 at a @xmath20-face . since @xmath412 ( @xmath413 ) , each @xmath414 misses exactly one vertex of @xmath58 , namely @xmath93 . this in turn implies that @xmath415 . but then the intersection of each @xmath414 with @xmath318 must necessarily be the unique ridge of @xmath318 containing @xmath416 , a contradiction . let @xmath48 be the ridge of @xmath2 containing @xmath417 . since @xmath37 is adjacent to every vertex in @xmath58 , @xmath48 must be a @xmath326-simplex . consider the other facet @xmath418 containing the ridge @xmath48 . note that every vertex in @xmath58 is simple in @xmath418 , and thus , @xmath37 must be simple in @xmath418 . let @xmath419 denote the unique edge incident with @xmath37 not contained in @xmath420 and let @xmath421 denote the set of facets containing @xmath36 . any facet @xmath422 must intersect @xmath317 at a @xmath20-face or at a ridge ; in any case , it must miss precisely one vertex of @xmath58 , say @xmath423 . in this case , @xmath29 intersects @xmath318 at a ridge , since @xmath60 and @xmath29 can not intersect @xmath318 at a @xmath20-face . it thus follows that for each @xmath423 there exists exactly one such facet @xmath29 . but there are only @xmath7 vertices in @xmath58 and we need @xmath8 facets in @xmath421 . this contradiction completes the proof of the lemma . [ lem : degree - d+1-ridges ] any semisimple @xmath1-polytope , in which every vertex has degree at most @xmath40 , is actually simple . we will show first that any facet @xmath29 is also semisimple . suppose otherwise ; that is there are two @xmath326-faces @xmath321 and @xmath322 in @xmath29 intersecting at a @xmath113-face @xmath424 . ( if @xmath424 had lower dimension , every vertex in @xmath424 would have excess at least 2 in @xmath29 . ) choose any @xmath425 . then @xmath16 has excess 1 in @xmath29 and so is nonsimple in @xmath2 ; [ rmk : simplenonsimplevertex ] ensures that @xmath16 is nonsimple in any facet containing it . denoting by @xmath426 the other facet containing @xmath338 ( @xmath413 ) , we see that @xmath16 is nonsimple in @xmath29 , @xmath317 and @xmath318 , and it has only one neighbour outside @xmath29 , so @xmath16 must be nonsimple in each @xmath338 . but then , the degree of @xmath16 in @xmath29 is the sum of the degree of @xmath16 in @xmath321 ( @xmath8 ) plus the edges of @xmath427 ( two ) , a contradiction . any 3-dimensional semisimple polytope must be simple by [ rmk : simplenonsimplevertex ] . the result now follows by induction on @xmath1 . [ lem : excess - d-1-ridges ] let @xmath2 be a semisimple @xmath1-polytope with excess @xmath8 . then @xmath24 , there is a unique nonsimple vertex in @xmath2 , and its vertex figure is combinatorially equivalent to @xmath150 . in particular , @xmath2 is a shephard polytope . recall that the excess is @xmath8 only if @xmath1 is odd . note that @xmath60 , since [ prop : charpsimple ] gives at once that all 3-dimensional semisimple polytopes are simple . by [ prop : charpsimple , thm : excess ] , we know that there exists a facet with excess at least @xmath27 , and by [ lem : facet - not - all - excess ] that the facet ( and every nonsimple facet ) has excess exactly @xmath27 ; note that since @xmath8 is even , the excess of a facet can not be @xmath7 . it follows that every nonsimple ridge has excess @xmath401 . we begin with the important observation that when a facet @xmath29 contains all the nonsimple vertices of @xmath2 , then @xmath2 contains at most two nonsimple vertices . from [ rmk : simplevertexoutside ] it follows that every nonsimple vertex in @xmath29 is adjacent to at least 2 vertices outside @xmath29 . the conclusion is now clear , since @xmath29 has excess @xmath27 and @xmath2 has excess @xmath8 . we consider first the case that @xmath2 contains a unique nonsimple vertex @xmath16 ; in fact , this is the only case which actually occurs . the vertex figure @xmath320 of @xmath2 at @xmath16 is a simple @xmath34-polytope with @xmath384 vertices ( cf . [ lem : vertex - all - neighbours - simple ] ) , from which [ lem : countsimplep](ii ) forces @xmath428 . since the case @xmath9 has been excluded , we have @xmath242 , and the vertex figure @xmath320 has @xmath385 vertices by virtue of [ lem : countsimplep ] . that is , @xmath320 is @xmath150 , which is depicted in [ fig : polytopes](c ) . henceforth we assume that @xmath2 has at least two nonsimple vertices . then any nonsimple vertex will be adjacent to at least one other nonsimple vertex . for , if a simple vertex were adjacent only to simple vertices , truncating it would give us a new simple facet with strictly less than @xmath181 vertices . next consider the case that @xmath2 contains exactly two nonsimple vertices , @xmath16 and @xmath37 . they must be connected by an edge , and hence contained in some facet @xmath29 . according to [ thm : excess - d-2-full - story ] , @xmath29 contains exactly @xmath27 vertices with excess one . so again @xmath396 . consider a facet @xmath429 containing @xmath16 but not @xmath37 , and a facet @xmath430 containing @xmath37 but not @xmath16 ( cf . [ rmk : two - vertices ] ) . since @xmath429 has excess @xmath396 , every vertex in @xmath429 sends exactly one edge outside @xmath429 . the vertex figure @xmath431 of @xmath429 at @xmath16 is a simple 3-polytope with six vertices ; that is , it is a simplicial 3-prism . this implies that the two 3-faces @xmath321 and @xmath322 of @xmath429 arising from the opposite simplices of @xmath431 intersect only at @xmath16 . consequently , the intersection of the other two facets of @xmath2 containing the ridges @xmath321 and @xmath322 has dimension at most one , a contradiction . now consider the case that @xmath2 contains exactly three nonsimple vertices , say @xmath432 and @xmath433 . without loss of generality , we can assume that @xmath434 is adjacent to both @xmath435 and @xmath433 . thanks to [ rmk : two - vertices ] , we can find a facet @xmath317 containing @xmath434 and @xmath433 but not @xmath435 , and a facet @xmath418 containing @xmath435 and @xmath434 but not @xmath433 . thanks to [ thm : excess - d-2-full - story ] , we have @xmath396 , each @xmath436 has excess 1 in the appropriate facet , and the total excess is 4 . clearly @xmath435 and @xmath433 can not both have excess 2 , so [ rmk : simplevertexoutside ] tells us that one of them is adjacent to a nonsimple vertex outside the corresponding facet . the only possibility is that @xmath435 and @xmath433 are also adjacent . we may assume that @xmath435 and @xmath433 have excess 1 , and that @xmath434 has excess 2 . let @xmath318 be a facet containing @xmath435 and @xmath433 but not @xmath434 , and denote the ridge @xmath437 by @xmath438 . then @xmath439 for each @xmath195 , @xmath436 has excess 1 in @xmath338 for @xmath440 , while @xmath322 is simple . now @xmath435 has 6 neighbouring vertices , and every facet containing it is nonsimple . it follows that @xmath435 is contained in exactly @xmath243 facets , one for each set of 5 of its neighbours . in particular , there are 4 facets containing @xmath432 and @xmath433 , contrary to our previous observation . if some facet contains a unique nonsimple vertex @xmath16 , then by [ thm : excess - d-2-full - story ] , the degree of @xmath16 in @xmath2 is at least @xmath27 , and there can be at most 3 nonsimple vertices in @xmath2 . this case was dealt with in the previous paragraph , so we now assume that every nonsimple facet @xmath29 contains @xmath27 nonsimple vertices , each with excess 1 in @xmath29 . again , there can be at most 2 nonsimple vertices outside @xmath29 . if @xmath2 contains exactly @xmath27 nonsimple vertices , they must all belong to the same facet , a situation whose impossibility has already been demonstrated . if @xmath2 contains exactly @xmath8 nonsimple vertices , they must all have excess 1 , and the impossibility of this is given by [ lem : degree - d+1-ridges ] . we are left with the case that @xmath2 contains exactly @xmath7 nonsimple vertices ; of these , one , say @xmath37 , will have excess 2 , while all the others have excess 1 . fix a vertex @xmath16 with excess 1 , and choose a facet @xmath29 containing @xmath16 but not @xmath37 . then @xmath29 must contain all @xmath27 vertices with excess 1 . the vertex @xmath16 has degree @xmath40 : its neighbours are @xmath37 , the other @xmath401 nonsimple vertices in @xmath29 , and four simple vertices in @xmath29 . in any facet which contains it , @xmath16 will be nonsimple and hence have @xmath1 neighbours . thus every such facet contains all but one of the nonsimple vertices , and all four simple vertices in @xmath29 . this leads to the absurdity that @xmath16 belongs to only @xmath27 facets . this impossibility completes the proof . [ thm : excessdecomp ] let @xmath2 be @xmath1-polytope with excess degree @xmath8 , where @xmath60 . then @xmath242 and either a. there is a single vertex with excess four , which is the intersection of three facets , and whose vertex figure is @xmath150 ; or b. there are two vertices with excess two , the edge joining them is the intersection of two facets , and its underfacet is either @xmath178 or @xmath179 ; or c. there are four vertices each with excess one , which form a quadrilateral 2-face which is the intersection of two facets , whose underfacet is the tesseract @xmath441 . in all cases @xmath2 is a shephard polytope . by [ lem : non - disjointfacets ] and [ rmk : excessd-1 ] any two nondisjoint facets in @xmath2 intersect at a vertex , an edge with @xmath242 , a subridge , or a ridge . applying in order [ lem : excessd-1facetsd-3,lem : excess4facets1,lem : excessd-1facets0 ] leads to cases ( i ) , ( ii ) and ( iii ) respectively . otherwise @xmath2 is semisimple , an impossibility according to [ lem : excess - d-1-ridges ] . in case ( i ) , the vertex figure is obviously @xmath150 . in case ( ii ) , the edge figure " must be a simple 4-polytope with 12 vertices , so by [ lem : countsimplep ] is either @xmath178 or @xmath179 . case ( iii ) is more complicated . let @xmath317 and @xmath318 be as in [ lem : excessd-1facetsd-3 ] . since we only need to calculate the face figure , we can assume that @xmath2 has a simple facet @xmath42 containing all vertices except the four in @xmath442 . each of these four is incident with two edges in @xmath317 and with two edges in @xmath318 . thus each ridge @xmath443 has 8 vertices , and is simple , so must be either a cube or a 5-wedge . however @xmath426 is the convex hull of @xmath444 , and is simple . if we remove an quadrilateral face from a copy of @xmath179 , the resulting graph is not the graph of a polytope . so for each @xmath195 , @xmath426 is a copy of @xmath178 , and @xmath338 is a cube . then @xmath29 , being the convex hull of @xmath321 and @xmath322 , and simple , must be a tesseract . examples for cases ( i ) , ( ii ) and ( iii ) are the pyramid over @xmath150 , @xmath296 and @xmath445 , and @xmath446 respectively . all of these have simple vertices , so repeated truncation leads to more examples . it is known ( * ? ? ? * theorem 7.1 , page 39 ) or ( * ? ? ? * theorem 9 ) that the only decomposable @xmath1-polytope with @xmath62 or fewer vertices is the prism . in this section , we characterise all the decomposable @xmath1-polytopes with @xmath12 vertices . characterising decomposable @xmath1-polytopes with @xmath22 vertices appears to be a much harder exercise . when @xmath9 , there are already 11 examples , namely @xcite . for @xmath160 , a brief discussion of this problem is given in the next section . recall that a @xmath1-dimensional capped prism is the convex hull of a simplicial prism and a single extra vertex , say @xmath90 , which lies beyond one of the simplex facets ( and beneath all the other facets ) . also recall that @xmath94 denotes the capped @xmath1-prism where @xmath67 is the minimum dimension of any face of the simplicial prism whose affine hull contains the _ extra _ vertex . if @xmath96 , the capped prism will be ( combinatorially ) just another prism . if @xmath98 , then @xmath2 is a pentasm , with @xmath100 edges . for @xmath196 , we can label the vertices as @xmath447 , @xmath448 in such a way that the edges , @xmath102 in total , are @xmath449 $ ] for all @xmath344 , @xmath450 $ ] for all @xmath344 , and @xmath451 $ ] for all @xmath452 . [ rmk : ] in the case of a pentasm , one edge , say @xmath453 $ ] will be absent , and then @xmath454 will form a pentagonal face . in the case of a prism , the vertex @xmath90 will be absent . [ twoout ] let @xmath2 be a decomposable @xmath1-polytope , @xmath29 a facet of @xmath2 , and suppose that there are only two vertices of @xmath2 outside @xmath29 . then @xmath29 is decomposable , and has shephard s property . the two vertices outside @xmath2 are not enough to form a facet . thus @xmath29 touches every facet , and so must be decomposable by [ thm : mcmullen - decomp ] . moreover if some vertex @xmath455 were adjacent to both vertices @xmath456 outside @xmath29 , then the triangle @xmath457 would be an indecomposable face touching every facet , contrary to the decomposability of @xmath2 . [ lowdim ] a. @xmath237 is not a facet of any decomposable 4-polytope with 9 vertices . b. @xmath150 is not a facet of any decomposable 5-polytope with 11 vertices . \(i ) let us entertain the possibility that @xmath2 is such a polytope , i.e. it has a facet @xmath29 of the type @xmath458 . by [ twoout ] , every vertex in @xmath29 belongs to only one edge not in @xmath29 , and thus @xmath2 has 19=11 + 7 + 1 edges . according to ( * ? ? ? * lemma 10(i ) ) , only the pentasm has 9 vertices and 19 edges , and it does not have @xmath237 as a facet ( cf . [ rmk : pentasm - facets ] ) . \(ii ) consider the possibility that @xmath2 is such a polytope , with @xmath150 as a facet , say @xmath29 . again , by [ twoout ] , every vertex in @xmath29 belongs to only one edge not in @xmath29 , and so @xmath2 has 28=18 + 9 + 1 edges . but then @xmath2 has excess degree only one , which is impossible ( cf . [ thm : excess ] ) . [ rmk : delta23 - 6polytope ] a similar argument proves that @xmath206 is not a facet of any decomposable 6-polytope with 14 vertices ; this will be useful to us in another context . [ twodeeplusone ] let @xmath2 be a decomposable @xmath1-polytope with @xmath12 vertices . a. every facet of @xmath2 with fewer than @xmath386 vertices is indecomposable . every facet of @xmath2 with exactly @xmath386 vertices is a prism . any facet of @xmath2 with @xmath384 vertices is decomposable , and moreover every vertex therein belongs to only one edge outside the facet . no facet of @xmath2 has @xmath62 vertices . b. if some decomposable facet of @xmath2 is a capped prism , then @xmath2 is a capped prism . c. if some decomposable facet of @xmath2 is a pentasm , then @xmath2 is a pentasm . d. if every decomposable facet of @xmath2 is a prism , then @xmath160 and @xmath2 is @xmath150 . \(i ) the first claim follows from ( * ? ? ? if a facet @xmath29 has @xmath386 vertices then there are only three other vertices in @xmath2 , which is not enough to form a facet . thus @xmath29 touches every facet , and must be decomposable by [ thm : mcmullen - decomp ] , hence a prism by ( * ? ? ? . 9 ) . for facets with @xmath384 vertices , this is just [ twoout ] . the last assertion follows from the indecomposability of pyramids . \(ii ) and ( iii ) . some facet @xmath29 is either a pentasm or a capped prism . we can write the vertex set of @xmath29 as @xmath459 , where @xmath460 , @xmath461 and the edges are as described above . note that the geometric subgraphs determined by @xmath462 and @xmath219 are indecomposable . let @xmath38 and @xmath463 be the vertices outside @xmath29 . we claim that one of them is adjacent only to vertices in @xmath462 , while the other is adjacent only to vertices in @xmath219 . suppose not . then one of them , say @xmath38 , is adjacent to both @xmath464 and @xmath465 for some @xmath344 . since @xmath466 , @xmath38 is adjacent to at least 3 vertices in @xmath29 . thus we may assume that @xmath467 . if @xmath29 is a capped prism , or if @xmath468 , then @xmath469 is a nonplanar 4-cycle sharing 2 vertices with @xmath219 . if @xmath470 , then @xmath471 is an affinely independent 5-cycle sharing 2 vertices with @xmath219 . then either @xmath472 or @xmath473 is an indecomposable subgraph containing all but @xmath8 or @xmath7 vertices of @xmath2 , hence touching every facet . this is inconsistent with the decomposability of @xmath2 by [ thm : mcmullen - decomp ] . so we have @xmath463 ( say ) adjacent to every @xmath465 , and @xmath38 adjacent to every @xmath474 . this describes the graph of @xmath2 completely ; it is the graph of either a pentasm or a capped prism . we know already that a pentasm is determined uniquely by its graph , since it is the unique @xmath1-polytope with @xmath12 and @xmath100 edges , except for @xmath9 where @xmath237 also share these properties ; see [ thm : pentasm ] . in case @xmath29 is a capped prism , a little more care is needed to determine the face lattice of @xmath2 , which we claim is also a capped prism ; we may also assume that no other facet is a pentasm . again , @xmath38 is adjacent to everything in @xmath462 , and @xmath463 to everything in @xmath219 , so it is reasonable to relabel @xmath475 and @xmath476 . thus @xmath2 has the same graph as a capped prism , but we need to reconstruct the whole face lattice . denote by @xmath477 the edge @xmath451 $ ] and by @xmath67 the minimum dimension of any face of the prism determined by @xmath478 whose affine hull contains @xmath90 . without loss of generality , we may suppose that @xmath90 lies in the affine hull of @xmath479 , and not to the affine hull of any smaller collection of these edges . for @xmath480 , let @xmath481 be the convex hull of @xmath482 , and let @xmath483 be the convex hull of @xmath484 . for @xmath485 , let @xmath481 be the convex hull of @xmath486 . it is not hard to see that each of these is a facet of @xmath2 , as is @xmath359 , the convex hull of @xmath487 . in particular , @xmath488 . we claim there are no other facets . since each @xmath464 is a simple vertex in @xmath2 , it must belong to exactly @xmath1 facets of @xmath2 . the list just given contains @xmath1 such facets for each @xmath464 . so any other facet must be contained in @xmath489 . all possible subsets are also accounted for by the list just given . we have now described the vertex - facet incidences of a capped prism . \(iv ) every decomposable facet is a prism ; fix one such facet @xmath29 . now there are three vertices outside @xmath29 ; call them @xmath490 . following the previous notation , @xmath29 is the convex hull of two simplices @xmath462 and @xmath219 ( both ridges in @xmath2 ) . note that if @xmath467 , then no vertex outside @xmath29 can be adjacent to both @xmath464 and @xmath465 . for if say @xmath491 were , then both @xmath492 and @xmath493 would be nonplanar 4-cycles with 3 vertices in common . their union with @xmath462 and @xmath219 would then be an indecomposable graph containing all but two vertices of @xmath2 . consider the possibility that one of the external vertices , say @xmath491 , is adjacent to both @xmath462 and @xmath219 . by the previous paragraph , there is a unique @xmath195 with @xmath491 adjacent to both @xmath464 and @xmath436 . now the degree of @xmath491 is at most 4 , so @xmath160 , @xmath29 is 3-dimensional and contains 3 quadrilateral and 2 triangular ridges . suppose @xmath42 is the other facet corresponding to one of the quadrilateral ridges . if @xmath42 were indecomposable , its union with the 2 triangular faces of @xmath29 would constitute an indecomposable graph containing at least 7 of the 9 vertices of @xmath2 . it follows that each of these 3 other facets must be a prism . this means that each of them contains only 2 of the vertices outside @xmath29 , each such vertex being adjacent to 2 vertices of the quadrilateral . this is only possible if @xmath494 can be renamed @xmath495 in such a way that they are adjacent to one another , and each @xmath496 is adjacent to both @xmath464 and @xmath436 . thus @xmath2 is simple and has the same graph as @xmath150 . the remaining case ( which turns out be impossible ) is that each vertex outside @xmath29 is adjacent only to @xmath462 or @xmath219 . without loss of generality , @xmath491 and @xmath497 are adjacent only to vertices in @xmath462 , and @xmath498 is adjacent to every vertex in @xmath219 . since @xmath89 , we can choose an index @xmath195 for which @xmath464 is adjacent to both @xmath491 and @xmath497 . we claim that @xmath498 can not be adjacent to both @xmath491 and @xmath497 . if it were , then @xmath499 and @xmath500 would both be 4-cycles in the graph of @xmath2 ; they can not both be coplanar . so ( at least ) one of them , say the first , is an indecomposable geometric graph , which shares two vertices with the graph determined by @xmath501 , and shares two vertices with the graph determined by @xmath502 . then union of these three graphs is indecomposable , and contains every vertex except @xmath497 . this would imply indecomposability of @xmath2 . without loss of generality , @xmath503 $ ] and @xmath504 $ ] are edges of @xmath2 , but @xmath505 $ ] is not . having degree ( at least ) @xmath1 , @xmath491 must then be adjacent to every vertex in @xmath462 . if @xmath497 is adjacent to only @xmath7 vertices in @xmath462 , we will have only @xmath100 edges altogether , and then @xmath2 would be a pentasm . otherwise , @xmath497 is also adjacent to every vertex in @xmath462 , and we have the same graph as a capped prism . either way , there are decomposable facets which are not prisms . [ thm : daviddecom2d+1 ] let @xmath2 be a decomposable @xmath1-polytope with @xmath12 vertices . then @xmath2 is either a pentasm , a capped prism , @xmath237 or @xmath150 . the case @xmath9 is easily checked via catalogues such as ( * ? ? ? 4 ) ; see @xcite for some examples of the reasoning . henceforth assume @xmath89 . for @xmath160 , suppose @xmath2 has 9 vertices and is decomposable . then it has a decomposable facet @xmath29 with at most 7 vertices , according to [ thm : d-1decomposablefacets ] . by [ lowdim](i ) , @xmath29 can not be @xmath237 , so it must be a prism , capped prism or pentasm . every possibility is covered by the various cases in [ twodeeplusone ] , so @xmath2 is either @xmath150 , a capped prism or pentasm . likewise if @xmath242 . by [ lowdim](ii ) , no decomposable facet @xmath29 can be @xmath150 , so the only options are prisms , capped prisms and pentasms . all these cases have been dealt with in [ twodeeplusone ] . finally , we can proceed with the induction . suppose @xmath506 , and that it has been established for every smaller dimension , and fix a @xmath1-dimensional decomposable polytope @xmath2 with @xmath12 vertices . then @xmath2 has decomposable facet @xmath29 , all with at most @xmath384 vertices . by induction , each such facet must be a prism , capped prism or pentasm , and [ twodeeplusone ] then completes the proof . the last proof incidentally proves that a conditionally decomposable @xmath1-polytope must have at least @xmath22 vertices . it also leads to the following . a polytope , whose graph is that of a capped prism , is a capped prism . let @xmath2 be such a @xmath1-polytope ; then @xmath2 has @xmath102 edges . we can label the vertices as @xmath447 , @xmath448 as before . we know that @xmath507 is simple , and that its neighbors are @xmath508 . it follows that one facet containing @xmath507 must contain @xmath509 but not @xmath90 . it is not hard to see that @xmath447 then form a facet , with shephard s property . since pentasms , @xmath237 and @xmath150 have strictly less than @xmath102 edges , @xmath2 must be a capped prism . recall that a 4-polytope with ten vertices must have at least 21 edges . we give the complete characterisation of 4-polytopes with ten vertices and precisely 21 or fewer edges ; we will need this in 7 . see [ fig:10v-21e ] for drawings of these polytopes . they are all all decomposable . we can also characterise the decomposable 4-polytopes with ten vertices and 22 edges . there are 6 such examples , as well as ( at least ) two indecomposable 4-polytopes with ten vertices and 22 edges ; we do not need the details of this characterisation here . more generally , we can characterise the polytopes with @xmath22 vertices and minimal number of edges in all dimensions , but details will appear elsewhere @xcite . [ thm : d=4 ] there are exactly four 4-polytopes with ten vertices and 21 or fewer edges : the polytopes @xmath357 , @xmath355 , @xmath510 and @xmath127 , all of which have the same @xmath104-vector @xmath511 . let @xmath2 be a 4-polytope with ten vertices and a minimum number of edges . according to ( * ? ? ? * ch . 10 ) or [ lem : countsimplep ] , @xmath2 is not simple , so we can assume it has exactly 21 edges . the excess degree of @xmath2 is exactly two , and consequently , any facet must have excess degree at most one thanks to [ lem : facet - not - all - excess ] . this implies that any facet with an even number of vertices must be simple , and with an odd number of vertices must have excess one . in particular , any facet with eight or fewer vertices can only be a cube , a 5-wedge , a pentasm , @xmath237 , a prism @xmath512 , a quadrilateral pyramid @xmath513 or a simplex . note that no facet can have nine vertices , because it would have at least @xmath514 edges , and @xmath2 , being a pyramid thereover would have at least 23 edges , contrary to hypothesis . we now distinguish several cases . * case 1 . * some facet @xmath29 is the cube . denote by @xmath16 and @xmath37 the two adjacent vertices outside @xmath29 . consider a quadrilateral @xmath321 of @xmath29 and the quadrilateral @xmath322 of @xmath29 opposite to @xmath321 . let @xmath317 and @xmath318 be the other facets containing @xmath321 and @xmath322 , respectively . if @xmath317 is a pyramid with apex @xmath16 , @xmath318 is another pyramid with apex @xmath37 . the face lattice of the resulting polytope is easily reconstructed from the graph ; it is @xmath357 ( cf . [ rmk : ad - facets ] ) . if no quadrilateral of @xmath29 is contained in a pyramid , then each quadrilateral must be contained in a simplicial 3-prism which contains both @xmath16 and @xmath37 . a moment s reflection shows that this is impossible . * some facet @xmath29 is a 5-wedge . first note that every vertex in @xmath29 has exactly one neighbour outside @xmath29 . denote by @xmath16 and @xmath37 the two adjacent vertices outside @xmath29 . consider the other facet @xmath515 containing one of the triangles . suppose it has five vertices . there are two 3-polytopes with five vertices but only one has excess degree one , the pyramid over a quadrilateral . but in this case a vertex in the triangle would be adjacent to both @xmath16 and @xmath37 , a contradiction . thus , we can assume @xmath515 is tetrahedron with apex @xmath16 . by analogy the other facet containing the other triangle in @xmath29 is also a tetrahedron , with apex @xmath37 . as in the previous case , the other facet corresponding to either quadrilateral face of @xmath29 must be a pyramid or a prism . it follows that the two remaining vertices on @xmath29 must be adjacent to the same vertex outside @xmath29 , say @xmath16 . thus , one quadrilateral is contained in a pyramid and the other is contained in a 3-prism . the face lattice can be reconstructed with no difficulty from the graph obtained ; the polytope is @xmath355 ( cf . [ rmk : bd - facets ] ) . * some facet @xmath29 is the 3-pentasm . first note that the other facet containing the pentagon ca nt be a pyramid ; otherwise the excess of @xmath2 would be at least three . if it has eight vertices ( i.e. is a 5-wedge ) , we can refer to case 2 . the remaining option is that it has seven vertices and must be another pentasm . between the two pentasms we have 17 edges and the remaining vertex outside @xmath29 , say @xmath16 , must be simple . consider the number @xmath516 of 2-faces in @xmath2 . the union of the pentasms gives 11 2-faces ( cf . [ rmk : pentasm - facets ] ) and the vertex @xmath16 is contained in @xmath517 2-faces . thus , @xmath518 . then euler s relation yields that the number of facets in @xmath2 is at most @xmath519 . but every @xmath1-polytope with @xmath64 facets is either a pyramid or simple ( * ? ? ? 6.1 ) , and @xmath2 is neither . * * some facet @xmath29 is @xmath237 . we may assume that no facet is a cube , 5-wedge or pentasm . denote by @xmath93 the vertex of excess degree one in @xmath29 . from [ thm : excess - d-2-full - story ] it ensues that @xmath2 is decomposable . let us fix a triangular ridge @xmath520 in @xmath29 and consider the other facet @xmath515 containing it . if @xmath515 were a quadrilateral pyramid , then every vertex outside @xmath29 would be simple in @xmath2 , and thus , every facet of @xmath2 would touch @xmath515 , implying by [ thm : mcmullen - decomp ] that @xmath2 is indecomposable , a contradiction . thus the other facets corresponding to triangular ridges in @xmath29 can only be simplices or prisms . if both are prisms , they contain a common triangle ( the three vertices outside @xmath29 ) and a common edge ( the one from @xmath93 running outside out of @xmath29 ) , meaning they must lie on the same 3-dimensional affine space . this is also impossible . suppose both are simplices . then one of the vertices outside @xmath29 will be adjacent to all three vertices in one the two triangles , another will be adjacent to all three vertices in the other triangle , and the third will be adjacent to both of the other vertices in @xmath29 . it is routine to verify that @xmath2 has the same face lattice as @xmath127 . in the remaining case , there must be exactly seven edges leaving @xmath29 and the three vertices outside @xmath29 are pairwise adjacent . we can denote one of them by @xmath491 , which sends three edges to @xmath29 , and the remaining two , by @xmath497 and @xmath498 , which each send two edges to @xmath29 . note that the vertex @xmath491 must be adjacent to the vertex @xmath93 by [ rmk : simplevertexoutside ] . the only possible pattern of connections between @xmath29 and the external vertices gives us the graph and face lattice of @xmath510 ( cf . [ rmk : cd - facets ] ) . * case 5 . * every facet is a prism , a quadrilateral pyramid or a simplex . by [ thm : excess - d-2-full - story ] no every facet can be simple . so it suffices to show that no facet can be a pyramid in this case . suppose some facet @xmath29 is a pyramid based on a quadrilateral ridge @xmath213 . denote by @xmath16 the apex of this pyramid and let @xmath42 be the other facet containing @xmath213 . then @xmath521 contains at most two vertices . if @xmath16 has degree six in @xmath2 , then every other vertex in @xmath2 is simple . if @xmath16 has degree five in @xmath2 , then the unique vertex adjacent to @xmath16 outside @xmath29 must be nonsimple ( cf . [ rmk : simplevertexoutside ] ) . in either case , every vertex in @xmath213 is simple in @xmath2 . thus all the neighbours of vertices of @xmath213 are in @xmath151 . if we remove the vertices in @xmath522 from the graph of @xmath2 , the resulting graph will be disconnected , contrary to balinski s theorem ( * ? ? ? * thm . 3.14 ) . we characterise all pairs @xmath13 for which there exists a 5-polytope with @xmath4 vertices and @xmath5 edges . in particular , we show that @xmath523 if @xmath4 is odd , and @xmath524 if @xmath4 is even and not eight . it is well known that @xmath525 . besides the simplex , there is a simple 5-polytope with @xmath4 vertices if , and only if , @xmath4 is even and @xmath526 . simplicity requires @xmath527 , so @xmath4 must be even . the simplex @xmath528 , the prism @xmath529 and our other friend @xmath206 are all simple , with 6 , 10 and 12 vertices respectively . truncating one vertex of any simple 5-polytope gives another simple 5-polytope with four more vertices . thus we obtain all even numbers from ten onwards . there is no example with eight vertices because @xmath530 . there is a nonsimple 5-polytope with @xmath4 vertices and @xmath5 edges , if , and only if , @xmath531 and @xmath532 . clearly we must have @xmath533 . any pair with @xmath534 or 2 is impossible by the excess theorem ( [ thm : excess ] ) . the case @xmath535 is also impossible by ( * ? ? ? * theorem 19 ) ; the proof of that result becomes somewhat shorter if we set @xmath242 . all other pairs except @xmath536 are possible . how to construct them ? the first step is to look at all 4-polytopes and construct pyramids over them . if a @xmath1-polytope has @xmath4 vertices and @xmath5 edges , then a pyramid thereover is a @xmath537-polytope with @xmath538 vertices and @xmath539 edges . grnbaum showed that @xmath540 $ ] , @xmath541 $ ] , @xmath542 $ ] , @xmath543 $ ] , @xmath544 $ ] , and @xmath545 $ ] for all @xmath546 . for @xmath547 , building pyramids on 4-polytopes with @xmath548 vertices shows that @xmath549 $ ] , @xmath550 $ ] , @xmath551 $ ] , @xmath552 $ ] , and @xmath553 $ ] . thus we have all the alleged examples for @xmath554 except @xmath555 and @xmath556 . but these are exemplified by the prism , the pentasm and the capped prisms respectively . note that these three examples all come from slicing a simple vertex off something else : a simplex , a triplex @xmath557 or a bipyramid over a 4-simplex , respectively . for @xmath558 , pyramids give all examples with @xmath559 . more precisely , for each such @xmath4 , we know that @xmath560 $ ] and hence @xmath561 $ ] . the cases @xmath562 require a little explanation . note that if @xmath563 , then a 5-polytope with @xmath4 vertices and @xmath5 edges must have at least one simple vertex . so suppose @xmath564 and @xmath562 . let @xmath67 be the smallest integer @xmath565 . then @xmath566 is either @xmath567 or @xmath568 . now put @xmath569 and @xmath570 . clearly @xmath571 , so there is a polytope with @xmath572 edges and @xmath573 vertices , at least one of which is simple , unless @xmath574 and @xmath575 . truncating a simple vertex @xmath67 times then gives a polytope with @xmath5 edges and @xmath4 vertices . in the case @xmath574 and @xmath575 , we have @xmath576 . the case @xmath577 comes from slicing a simple edge off a capped prism . all remaining cases with more than 17 vertices then come from repeated truncation of simple vertices . it only remains to prove the unfeasibility of @xmath578 . this has a lengthy proof , to be given separately . we state this last impossibility result here , but postpone the proof of this theorem to [ sec:13n-35e ] . [ thm : v13-e35 ] there is no 5-polytope with @xmath579 vertices and 35 edges . 99 d. w. barnette , _ a proof of the lower bound conjecture for convex polytopes _ , pacific j. math . * 46 * ( 1973 ) , 349354 . d. britton and j. d. dunitz , _ a complete catalogue of polyhedra with eight or fewer vertices _ , acta crystallographica section a * 29 * ( 1973 ) , no . 4 , 362371 . j. federico , _ polyhedra with @xmath391 to @xmath580 faces _ , geometriae dedicata * 3 * ( 1974/75 ) , 469481 . b. grnbaum , _ polytopes , graphs , and complexes _ , bull . * 76 * ( 1970 ) , 11311201 . b. grnbaum , _ convex polytopes _ , 2nd ed . , graduate texts in mathematics , vol . 221 , springer - verlag , new york , 2003 , prepared and with a preface by v. kaibel , v. klee and g. m. ziegler . g. kalai , _ rigidity and the lower bound theorem . i _ , invent . math . * 88 * ( 1987 ) , no . 1 , 125151 . m. kallay , _ decomposability of convex polytopes _ , ph.d . dissertation , the hebrew university of jerusalem , 1979 . m. kallay , _ indecomposable polytopes _ , israel j. math . * 41 * ( 1982 ) , no . 3 , 235243 . kirkman , _ applications of the theory of the polyedra to the enumeration and registration of results _ , proc . royal soc . london * 12 * ( 1862 - 3 ) , 341380 . victor klee , _ polytope pairs and their relationship to linear programming _ , acta math . * 133 * ( 1974 ) , 125 . victor klee , _ adjoints of projective transformations and face - figures of convex polytopes _ , math . programming stud . * 8 * ( 1978 ) , 208216 . victor klee and david w. walkup , _ the @xmath1-step conjecture for polyhedra of dimension @xmath581 _ , acta math . * 117 * ( 1967 ) , 5378 . takuya kusunoki and satoshi murai , _ the numbers of edges of 5-polytopes _ , arxiv:1702.06281 , 2017 . c. w. lee , _ counting the faces of simplicial convex polytopes _ , thesis , cornell university , jan 1981 , p. 180 . p. mcmullen , _ the minimum number of facets of a convex polytope _ , j. london math . ( 2 ) * 3 * ( 1971 ) , 350354 . g. pineda - villavicencio , j. ugon , and d. yost , _ lower bound theorems for general polytopes _ , arxiv:1509.08218 , 2015 . , _ characterisation of d - polytopes with 2d+2 vertices and minimum number of edges _ , in preparation , 2016 . k. przesawski and d. yost , _ decomposability of polytopes _ , discrete comput . geom . * 39 * ( 2008 ) , no . 1 - 3 , 460468 . , _ more indecomposable polyhedra _ , extracta mathematicae * 31 * ( 2016 ) , no . 2 , 169188 . g. c. shephard , _ decomposable convex polyhedra _ , mathematika * 10 * ( 1963 ) , 8995 . z. smilansky , _ decomposability of polytopes and polyhedra _ , geom . dedicata * 24 * ( 1987 ) , no . 1 , 2949 . d. yost , _ irreducible convex sets _ , mathematika * 38 * ( 1991 ) , no . 1 , 134155 . g. m. ziegler , _ lectures on polytopes _ , graduate texts in mathematics , vol . 152 , springer - verlag , new york , 1995 . this appendix completes the proof of [ thm : v13-e35 ] . we proceed by examining the numbers of vertices in the facets of such a hypothetical polytope . the proof of kusunoki and murai @xcite , is quite different , considering instead the possible degrees of the vertices of the polytope . we first require some lemmas . note that the excess of @xmath2 is five . suppose @xmath29 is a facet of @xmath2 with ten vertices ; then @xmath29 has 22 or 21 edges . let @xmath38 , @xmath463 and @xmath582 denote the vertices in @xmath583 . we can assume that @xmath38 and @xmath463 , and @xmath38 and @xmath582 are adjacent . furthermore , every other facet of @xmath2 intersects @xmath29 at a ridge ; otherwise there would be at least thirteen edges between @xmath29 and @xmath584 , giving a minimum of 36 edges of in @xmath2 . this property implies that the number of facets other than @xmath29 in @xmath2 coincides with the number of 3-faces of @xmath29 , and that every vertex simple in @xmath29 is simple in @xmath2 . first suppose that @xmath463 and @xmath582 are not adjacent . then there is a unique vertex @xmath16 in @xmath29 sending two edges outside @xmath29 . the vertices @xmath38 , @xmath463 and @xmath582 are all simple . if there were another nonsimple vertex in @xmath29 , then , by [ rmk : simplevertexoutside ] , the vertex would have two neighbours outside @xmath29 . consequently , the vertex @xmath16 is the unique nonsimple vertex of @xmath29 ( and of @xmath2 ) , with degree ten in @xmath2 . but then , [ lem : vertex - all - neighbours - simple ] would imply that the vertex figure of @xmath2 at @xmath16 is a simple 4-polytope with ten vertices , which is ruled out by [ lem : countsimplep ] . now suppose that @xmath463 and @xmath582 are adjacent , with @xmath463 having degree six in @xmath2 . in this case , there are ten edges leaving @xmath29 . since the vertices @xmath38 and @xmath582 are simple , there is a facet @xmath42 in @xmath2 containing @xmath38 and @xmath582 but not @xmath463 . by [ rmk : simplevertexoutside ] the six vertices in @xmath585 are simple in @xmath2 . the ridge @xmath48 must therefore be a simplicial 3-prism , according to [ lem : countsimplep ] . looking at the structure of @xmath29 , we see that nine edges of @xmath29 come from @xmath48 and that there are six edges between @xmath48 and the four vertices in @xmath49 . but this counting leaves @xmath586 edges to be distributed among the four vertices in @xmath49 , a contradiction . there are 14 edges and three vertices @xmath38 , @xmath463 and @xmath582 outside @xmath29 . we can assume that @xmath38 and @xmath463 , and @xmath38 and @xmath582 are adjacent . furthermore , @xmath29 is one of the four polytopes presented in [ thm : d=4 ] , all of which has @xmath104-vector @xmath511 . suppose that @xmath463 and @xmath582 are not adjacent . without loss of generality , there are two possible configurations : either the vertex @xmath38 or the vertex @xmath463 has degree six in @xmath2 . we estimate the number of 3-faces of @xmath29 . in the first configuration , there are at least six facets containing @xmath38 , since @xmath38 is nonsimple . and there must be two further facets in @xmath2 , one containing @xmath463 but not @xmath38 or @xmath582 , and another one containing @xmath582 but not @xmath38 or @xmath463 , which contradicts the seven 3-faces of @xmath29 . analogously , in the second configuration , there are at least six facets containing @xmath463 , since @xmath463 is nonsimple . and there must be two further facets in @xmath2 , one containing @xmath38 and @xmath582 but not @xmath463 , and another one containing @xmath582 but not @xmath38 or @xmath463 , which contradicts the seven 3-faces of @xmath29 . suppose that @xmath463 and @xmath582 are adjacent . then there are exactly eleven edges leaving @xmath29 . since every vertex simple in @xmath29 is simple in @xmath2 , there must be a unique nonsimple vertex , say @xmath16 , in @xmath29 sending two edges outside @xmath29 . without loss of generality , there are two possible configurations : either the vertices @xmath38 and @xmath463 have both degree 6 in @xmath2 or the vertex @xmath463 has degree seven in @xmath2 . we first prove the following claim . refer to [ fig:10v-21e](a ) and [ rmk : ad - facets ] . consider both configurations mentioned above . the other nonsimple vertex @xmath37 in @xmath29 must be adjacent to a nonsimple vertex outside @xmath29 , say @xmath463 , by [ rmk : simplevertexoutside ] . this in turn implies that @xmath463 is contained in each of the five facets obtained from the five 3-faces in @xmath29 containing @xmath37 . the sixth facet containing @xmath463 must then be the one obtained from the 3-cube . consequently , the other facet containing the quadrilateral pyramid with @xmath16 as apex must contain @xmath38 and @xmath582 but not @xmath463 , and it must be a @xmath587-triplex ( cf . [ thm : triplexes ] ) , with both @xmath38 and @xmath582 adjacent to @xmath16 . without loss of generality , assume the vertex @xmath582 is adjacent to @xmath588 and @xmath589 , and @xmath38 is adjacent to @xmath590 and @xmath591 . this is in turn implies that @xmath463 is adjacent to at least three vertices in @xmath592 and @xmath38 is adjacent to at most one vertex in @xmath592 . in the first configuration , the other facet @xmath42 containing the quadrilateral pyramid @xmath593 contains @xmath38 and @xmath463 but not @xmath582 , with @xmath38 only having degree two in @xmath42 , a contradiction . in either configuration , the edge @xmath594 must be contained in exactly four facets containing @xmath16 " those obtained from 3-faces of @xmath29 containing @xmath16 ; three of such facets must come from simplicial 3-prisms in @xmath29 , with at least one of them , say @xmath213 , also containing the vertices @xmath38 and @xmath463 . observe that the fifth facet containing @xmath582 must be the one obtained from the 3-cube . for the second configuration , the facet @xmath42 is a pyramid over @xmath593 , and without loss of generality , we can assume that the facet @xmath213 contains the simplicial 3-prism @xmath595 . in this case , @xmath213 would have nine vertices , 19 edges , and two nonsimple vertices @xmath16 and @xmath463 which are not adjacent . from [ thm : pentasm ] it ensues that @xmath213 would be the pentasm , but the two nonsimple vertices of the 4-pentasm are adjacent . this contradiction concludes the proof of the claim . in the first configuration , since @xmath582 is simple , there are exactly five facets containing @xmath582 , each containing @xmath38 or @xmath463 . since @xmath38 and @xmath463 are nonsimple and @xmath29 has exactly seven 3-faces , every facet containing @xmath38 but not @xmath582 must contain @xmath463 , and the same for @xmath463 ; otherwise , to guarantee six facets for each @xmath38 and @xmath463 , apart from the facets containing @xmath582 , there would be at least three more facets : one containing @xmath38 but not @xmath582 or @xmath463 , one containing @xmath38 and @xmath463 but not @xmath582 , and one containing @xmath463 but not @xmath38 or @xmath582 . thus @xmath29 must be @xmath357 , as it does nt contain simplices as 3-faces , which in turn contradicts claim 1 . in the second configuration , there are exactly five facets containing @xmath38 , since @xmath38 is simple . each of the facets containing @xmath38 contains a 3-face of @xmath29 with at least five vertices . since @xmath582 is simple , there is exactly one facet containing @xmath582 but not @xmath38 ; this facet contains @xmath463 and a 3-face of @xmath29 with at least five vertices . thus we have already counted six 3-faces of @xmath29 with five or more vertices , so @xmath29 can not be @xmath355 or @xmath127 as they both have at most five 3-faces with five or more vertices ; see [ rmk : bd - facets , rmk : sigmad - facets ] . moreover , from claim 1 it follows that @xmath29 is @xmath510 . refer to [ fig:10v-21e](c ) and [ rmk : cd - facets ] . the other nonsimple vertex @xmath37 in @xmath29 must be adjacent to the nonsimple vertex @xmath463 outside @xmath29 , by [ rmk : simplevertexoutside ] . this in turn implies that @xmath463 is contained in each of the five facets obtained from the five 3-faces in @xmath29 containing @xmath37 . the sixth facet containing @xmath463 must then be the one obtained from the copy of @xmath237 which does not contain @xmath37 : @xmath596 . let @xmath42 be the other facet containing the simplicial 3-prism which does not contain @xmath16 or @xmath37 : @xmath597 . then the vertices @xmath38 and @xmath582 send each three edges to @xmath42 , and @xmath463 sends its five edges into the four vertices of the simplex of @xmath510 : @xmath598 . this is a contradiction because the other facet containing the simplex must be another simplex . this final contradiction rules out this configuration , and completes the proof of the lemma . suppose otherwise . since the excess of @xmath2 is two , [ thm : excess - d-2-full - story ] tells us that it is decomposable and has a facet with excess one . moreover , [ lem : facet - not - all - excess ] tells us that no facet has excess two , so every facet of @xmath2 with an odd number of vertices has excess one . if @xmath2 were a pyramid , it would have at least @xmath599 edges . if instead @xmath2 had a 3-face @xmath29 with nine vertices , @xmath29 would have more than @xmath600 edges and @xmath2 would have at least @xmath601 edges . thus the largest 3-face in @xmath2 has at most seven vertices . suppose there is a 3-face @xmath29 with seven vertices . then @xmath29 must have 11 edges , and by [ thm : pentasm ] is a 3-pentasm . denote by @xmath602 the number of edges among the four vertices outside @xmath29 and by @xmath603 the number of edges between @xmath29 and the four vertices in @xmath39 . then @xmath604 , @xmath605 and @xmath606 . this implies that @xmath607 . in both cases , the other 3-face of @xmath2 containing the pentagon in @xmath48 must contain at least three of the vertices in @xmath39 , giving a facet with at least eight vertices , a contradiction . finally , suppose there is a facet @xmath29 with at five vertices . then it has excess one and is a quadrilateral pyramid with apex @xmath16 ( cf . [ thm : triplexes ] ) . suppose there is a nonsimple vertex @xmath37 in @xmath39 . then , the degrees of @xmath16 and @xmath37 in @xmath2 are five , @xmath16 and @xmath37 are adjacent by [ rmk : simplevertexoutside ] , and there are exactly five edges between @xmath29 and @xmath39 . consider the other facet @xmath42 containing the quadrilateral @xmath213 of @xmath29 . it follows that either @xmath42 is a quadrilateral pyramid with apex @xmath37 or a simplicial 3-prism . but then , removing either @xmath16 and @xmath37 in the former case , or @xmath16 and the two vertices @xmath608 in the latter case would disconnect @xmath213 , contradicting balinski s theorem ( * ? ? ? * thm . 3.14 ) . if instead every vertex in @xmath39 is simple . then , the degrees of @xmath16 @xmath2 are six @xmath16 and there are exactly six edges between @xmath29 and @xmath39 . again , considering the other facet @xmath42 containing the quadrilateral @xmath213 of @xmath29 , it follows that @xmath42 is a simplicial 3-prism . but then , removing @xmath16 and the two vertices @xmath608 would disconnect @xmath213 , contradicting balinski s theorem ( * ? ? ? 3.14 ) . let @xmath29 denote a facet of @xmath2 with 11 vertices . the facet can not have more than 23 edges , since there must be at least twelve edges outside @xmath29 . thus the facet @xmath29 has 22 or 23 edges . in either case , any other facet @xmath42 of @xmath2 intersects @xmath29 at a ridge ; indeed , any intersection of smaller dimension would require at least thirteen edges to leave @xmath29 . we consider each subcase at a time . clearly the excess of @xmath29 can only be 0 , 2 or 4 . from @xcite , it follows that @xmath29 can not have 17 edges , i.e. its excess can not equal 2 ( see ( * ? ? ? 19 ) for a higher dimensional version of this ) . so suppose that @xmath29 has excess 4 , i.e. there are 18 edges in @xmath29 . then @xmath29 is indecomposable by ( * ? ? ? 9 ) , and the five vertices @xmath609 outside @xmath29 can not form a facet , as @xmath2 would then have at least 36 edges . hence the whole polytope is indecomposable by [ thm : mcmullen - decomp ] , and thus @xmath29 can not have shephard s property by [ thm : shp ] . it follows that there are 9 edges running out of @xmath29 , 8 edges joining the @xmath610 , each @xmath610 is simple . however every nonsimple vertex of @xmath2 which is contained in @xmath29 must send two edges outside @xmath29 , which means there would be a unique nonsimple vertex in @xmath2 ( actually in @xmath29 ) , which can only have degree ten in @xmath2 . but then , [ lem : vertex - all - neighbours - simple ] gives that the corresponding vertex figure of @xmath2 would be a simple 4-polytope with 10 vertices , which is ruled out by [ lem : countsimplep ] . let @xmath2 denote a @xmath390-polytope with 13 vertices and 35 edges . throughout , @xmath29 will denote a fixed facet with the maximal number of vertices . we investigate various cases , based on this number . from [ lem : facet - not - all - excess ] it ensues that the facet @xmath29 has excess zero ( 18 edges ) , excess two ( 19 edges ) or excess four ( 20 edges ) . denote by @xmath334 , @xmath335 , @xmath336 and @xmath337 the vertices outside @xmath29 . we consider each subcase at a time . if every vertex in @xmath39 were simple in @xmath2 , then by [ rmk : simplevertexoutside ] every nonsimple vertex of @xmath2 which is contained in @xmath29 would send two edges outside @xmath29 . in this case , there would be a unique nonsimple vertex in @xmath29 ( and in @xmath2 ) , say @xmath16 , with degree ten in @xmath2 . but then , [ lem : vertex - all - neighbours - simple ] gives that the vertex figure of @xmath2 at @xmath16 would be a simple 4-polytope with ten vertices , which is ruled out by [ lem : countsimplep ] . consequently , there is a unique nonsimple vertex in @xmath39 , say @xmath334 , which has degree six in @xmath2 and is adjacent to to every nonsimple vertex lying in @xmath29 . furthermore , there are exactly nine edges between @xmath39 and @xmath29 , and exactly six edges shared among the vertices @xmath334 , @xmath335 , @xmath336 and @xmath337 . we count the number of facets involving the vertices @xmath334 , @xmath335 , @xmath336 and @xmath337 . there are exactly five facets containing @xmath336 , since @xmath336 is simple . since @xmath337 is simple , there is exactly one facet containing @xmath337 but not @xmath336 ; this facet also contains @xmath334 and @xmath335 . in this way , we have counted all the facets containing the vertices @xmath335 , @xmath336 and @xmath337 . out of of the aforementioned six facets , the nonsimple vertex @xmath334 can not be contained in all of them , since there must exist a facet containing @xmath336 but not @xmath334 ( cf . [ rmk : two - vertices ] ) . but then the nonsimple vertex @xmath334 is only contained in five facets , a contradiction . then @xmath29 is a @xmath391-pentasm . there is 3-face @xmath48 of @xmath29 which is a 3-pentasm ( cf . [ rmk : pentasm - facets ] ) , i.e. it has 7 vertices and 11 edges . denote by @xmath42 the other facet containing @xmath48 . we know that @xmath42 can not have 10 or more vertices , and it can not be @xmath150 because it contains a pentagonal face . so if it has 9 vertices , then it has at least 19 edges . but then @xmath151 contains at least 27 edges , while the two vertices outside @xmath151 must contribute at least 9 more edges . finally , consider the possibility that @xmath42 contains 6 vertices . then @xmath42 is a pyramid over @xmath48 , @xmath151 contains 25 edges , and the three vertices outside @xmath151 must contribute at least 12 more edges . [ thm : pentasm ] tells us that @xmath29 is @xmath150 , and thus , has six 3-faces , all of them being simplicial 3-prisms ( cf . [ fig : polytopes](c ) ) . there are 17 edges outside @xmath29 . denote the vertices outside by @xmath334 , @xmath335 , @xmath336 and @xmath337 . any subset of three of these vertices will belong to at least 12 edges ; hence any one of them can be adjacent to at most five vertices in @xmath29 . it follows that for any 3-face @xmath48 in @xmath29 , the corresponding other facet is never a pyramid over @xmath48 . this in turns implies that each @xmath610 is adjacent to at least two of the others ( if there were say only one edge from @xmath334 to the other @xmath610 , the only facet containing @xmath334 but not this edge would be a pyramid with apex @xmath334 and base in @xmath29 ) . for any ridge @xmath48 in @xmath29 , the corresponding other facet must have either 8 or 9 vertices , and so must be either a 4-prism or a copy of @xmath150 . so if @xmath520 is any triangular face in @xmath29 , any one of the external vertices @xmath610 must be adjacent to either 0 , 1 or 3 of the vertices in @xmath520 ; 2 is not possible . recall that the 3-faces in @xmath150 can be partitioned into two groups of three ; any two 3-faces within the same group intersect in a triangle , and any two 3-faces from different groups intersect in a quadrilateral . the structure of the other facets corresponding to ridges in @xmath29 is then essentially determined . without loss of generality we can assume that for @xmath612 , @xmath610 is adjacent to @xmath613 , @xmath614 and @xmath615 , but not to @xmath616 , @xmath617 or @xmath618 for @xmath619 , and that @xmath334 , @xmath335 , @xmath336 are mutually adjacent . three of these facets are 4-prisms , three of them are copies of @xmath150 , and none of them contain @xmath337 . the last assertion will lead to an absurdity . suppose @xmath228 is any facet containing @xmath337 . clearly @xmath228 intersects @xmath29 at a non - empty face of dimension @xmath620 . every vertex in @xmath29 is adjacent to precisely one of @xmath334 , @xmath335 , @xmath336 ; this rules out the possibility that @xmath621 is a single vertex or an edge . were @xmath621 a quadrilateral , each of its vertices would be adjacent to two vertices outside @xmath29 , and there would be at least 13 edges running out from @xmath29 . but then @xmath337 could only be adjacent to one of the other @xmath610 , contrary to our previous conclusion . thus @xmath622 must be a triangle . now let @xmath229 be another facet containing @xmath337 , but not containing @xmath623 . then @xmath624 is another triangle , each of whose vertices is also adjacent to two vertices outside @xmath29 . thus there would be at least 10 edges running out from @xmath29 from @xmath625 , and another 4 from the remaining vertices . but then @xmath337 could not be adjacent to any of the other @xmath610 . observe that every two simplicial 3-prisms in @xmath29 intersect in a quadrilateral 2-face . consider one of the simplicial 3-prisms @xmath321 in @xmath29 . if the other facet @xmath317 containing @xmath321 is a pyramid , then @xmath626 contains 22 edges and there are four vertices outside @xmath626 incident to 13 edges ; this is impossible since four vertices are incident to at least 14 edges . so @xmath317 ( and every facet containing of one of the simplicial 3-prisms in @xmath29 ) must be a simplicial prism . let @xmath334 and @xmath335 denote the vertices in @xmath627 . consequently , there are 12 edges outside @xmath626 incident to the remaining three vertices @xmath336 , @xmath337 and @xmath628 . so every pair in @xmath629 is adjacent , each vertex in @xmath629 is simple , and at least one vertex in @xmath629 is adjacent to a vertex in @xmath630 . this in turn implies that every vertex @xmath631 sends at most three edges into @xmath29 . the facet @xmath29 has four different simplicial 3-prisms , say @xmath321 , @xmath322 , @xmath340 and @xmath341 , and correspondingly , four different simplicial 4-prisms @xmath317 , @xmath318 , @xmath418 and @xmath632 containing these 3-faces . if a vertex in @xmath631 were contained in more than one of the facets @xmath317 , @xmath318 , @xmath418 and @xmath632 , it would send at least four edges into @xmath29 , a contradiction . consequently , the pairs of vertices in @xmath633 must be pairwise disjoint , which is clearly a contradiction . the minimum number of edges of @xmath29 is 15 ( cf . [ thm : triplexes ] ) , in which case @xmath634 , that is , a pyramid over a simplicial 3-prism @xmath48 . suppose @xmath29 has 15 edges and consider the other facet @xmath317 containing @xmath48 . then @xmath635 , and @xmath636 contains 21 edges . hence there are 14 edges outside @xmath636 which are incident to five vertices of degree at least 5 , which is impossible . thus @xmath29 has at least 16 edges . if @xmath29 has exactly 16 edges , then , analysing ( * ? ? ? * fig . 5 ) we get that there are exactly two such polytopes , each being a pyramid over a 3-polytope with six vertices and ten edges ; see ( * ? ? ? 3 ) for a description of the 3-polytopes . as before , let @xmath48 be the 3-face with 10 edges in @xmath29 . the other facet @xmath317 containing @xmath48 is also a pyramid , and thus , and @xmath636 contains 22 edges . hence there are 13 edges outside @xmath636 which are incident to five vertices of degree at least five , which is again impossible . the facet @xmath29 has at least 17 edges . denote by @xmath603 the number of edges between the the six vertices outside @xmath29 and the facet @xmath29 , and by @xmath602 the number of edges among the six vertices outside @xmath29 . then @xmath606 . on the other hand , @xmath637 and @xmath638 , implying @xmath639 and @xmath640 , a contradiction . the minimum number of edges of @xmath29 is 13 ( cf . [ thm : triplexes ] ) , in which case @xmath641 , that is a two - fold pyramid over a quadrilateral . suppose @xmath29 has 13 edges and consider the other facet @xmath317 containing the pyramid @xmath48 over a quadrilateral . then @xmath642 , and @xmath636 contains 18 edges . hence there are 17 edges outside @xmath636 which are incident to six vertices of degree at least 5 . let @xmath359 denote the set of the six vertices outside @xmath636 . denote by @xmath603 the number of edges between @xmath359 and @xmath636 , and by @xmath602 the number of edges among the six vertices in @xmath359 . then @xmath643 and @xmath638 , implying @xmath644 and @xmath645 . however , having @xmath645 contradicts balinski s theorem ( * ? ? ? * thm . 3.14 ) , since removing at most four vertices in @xmath359 , those incident to the edges counted in @xmath603 , disconnects the polytope graph . thus @xmath29 has at least 14 edges . from ( * ? ? ? 6.1.4 ) it ensues that there are exactly three 4-polytopes with 6 vertices and at least 14 edges , all of them being 2-simplicial , i.e. every @xmath367-face is a simplex . this implies that @xmath2 is 2-simplicial , and therefore ( * ? ? ? * thm . 1.4 ) ) gives the final contradiction as it states that @xmath2 would have at least 50 edges .	we define the excess degree @xmath0 of a @xmath1-polytope @xmath2 as @xmath3 , where @xmath4 and @xmath5 denote the number of vertices and edges , respectively . we first prove that the excess degree of a @xmath1-polytope does not take every natural number : the smallest possible values are @xmath6 and @xmath7 , and the value @xmath8 only occurs when @xmath9 or 5 . on the other hand , for fixed @xmath1 , the number of values not taken by the excess degree is finite ( subject to the restraint that if @xmath1 is even , then the excess degree must be even ) . our study of the excess degree is then applied in three different settings . we show that polytopes with small excess ( i.e. , @xmath10 ) behave in a similar manner to simple polytopes in terms of minkowski decomposability : each such polytope is either decomposable or a pyramid , and their duals are always indecomposable . ( this is no longer true when @xmath11 . ) secondly , we characterise the decomposable @xmath1-polytopes with @xmath12 vertices . and thirdly , we characterise all pairs @xmath13 for which there exists a 5-polytope with @xmath4 vertices and @xmath5 edges .
in depth filtration , suspended particles in a fluid are removed during their passage through a porous medium@xcite . the basic dynamics of depth filtration is determined primarily by the pore structure of the filter , the particle size distribution , and by various physicochemical and hydrodynamic details . if the particle size is larger than the typical pore size , particles get stuck relatively quickly . the permeability of the filter decreases steadily during this process and drops to zero when clogging is reached . this process is often referred to as sieving , or straining@xcite . conversely , if particles are much smaller than the pore size and if particles are trapped only at the interfaces of the porous medium , the flow field is only slightly affected by the trapping . the goal of this paper is to provide a general understanding of this latter process of _ infiltration _ by microscopic network modeling . 2.0 cm .4 cm infiltration underlies many practical situations , such as underground waste disposal @xcite , gas mask design , or drinking water filters @xcite . typically , sub - micron size contaminant particles are suspended in a carrier fluid and flow through a porous material , such as a sand filter whose typical grain size is much larger than the contaminant particles , or an ion exchange filter@xcite where the contaminant size is molecular in scale . in such cases , one can neglect the change of the flow field due to particle trapping @xcite , an approximation which considerably simplifies theoretical analysis . the kinetics of infiltration is controlled by the microscopic mechanisms for the trapping of the invader particles . typically each pore can hold a limited number of particles due to a finite surface area or a finite range of the surface potential . when all the available surface area is covered by particles , subsequent invaders flow passively through the filter without being trapped . our basic goal is to understand the kinetics of this infiltration and the ultimate breakthrough of the invader , as well as the evolution of the invader and defender density profiles as functions of downstream position and time . previous work on infiltration in porous media has often been based on a macroscopic convection - diffusion equation description , with reaction terms introduced to account for particle trapping @xcite . another approach has been to use a single absorbing sphere to calculate the collection efficiency at the initial stage of filtration@xcite . while numerical simulations of these models have some predictive power , it is hard to develop a connection between this macroscopic approach and basic features of the microscopic process , such as the concentration profiles of the trapped and flowing particles . for filtration by straining , models based on a discrete network description of the filter medium are relatively well developed @xcite . to our knowledge , however , there has been no microscopic network modeling work on infiltration . as in the case of straining , a spatial density gradient naturally arises in infiltration , since particles begin to deposit at the upstream end of the filter and advance downstream as the filter gets used up . the density gradient is experimentally observed as the time - dependent output concentration @xcite . in this paper , we will account for this basic experimental observation by using a discrete network approach . practical questions raised by infiltration are the breakthrough time , which is defined as the time for the output concentration to reach a specified threshold level , and the filter efficiency , which is related to the fraction of the filter material actually used before breakthrough . clearly , it is desirable to use as much of the filter material as possible before breakthrough occurs . this paper is organized as follows . in sec . [ basic ] , we introduce the basic parameters that govern particle trapping and provide a qualitative picture of infiltration . in the following sections , we construct a sequence of discrete models with increasing complexity and realism to ultimately provide a lattice network description . in sec . [ 1d ] , we discuss the case of a one - dimensional ( 1d ) chain of trapping sites and in sec . [ bubble ] , we analyze infiltration in the bubble model to provide a mean - field - like description . building on these results , we then turn to simulations of infiltration on tube lattice networks in sec . [ network ] . we summarize and compare our results with experiments in sec . [ discussion ] . the two basic characteristics of particle trapping are the efficiency of an unoccupied trapping site and the number of trapping sites in a pore . we introduce the trapping probability @xmath0 as the probability that a particle is trapped upon encountering an open collector site . the parameter @xmath0 thus represents the strength of the particle - collector interaction and accounts for the possibility that contact between particles and the filter grains may not necessarily lead to deposition @xcite . while this simplifies the complicated adsorption mechanism , later we show that the basic feature such as the invasion front propagation velocity ( fig.[fig1 ] ) is independent of the interaction details . next we introduce the capacity @xmath1 as the number of particles a pore at position @xmath2 can hold at time @xmath3 . in the case of non - coagulating particles which can not get trapped on top of an already adsorbed particle , the initial capacity is proportional to the inner surface area of the pore and then decreases as the pore surface is covered by particles . for simplicity , we ignore multiple trapping on an already occupied collector site as well as particle re - launching . the key factors which determine the dynamic behavior of the system are geometric , such as the capacity of a clean filter and the pore size distribution , and kinematic , such as the particle concentration and the flow rate . more refined models for particle trapping can be incorporated within our basic modeling . consider a generic infiltration process based on the above concepts . initially a layer in the clean filter has a total capacity @xmath4 independent of the downstream position . at @xmath5 , a fluid which contains a mixture of invader and non - reacting tracer particles enters the filter whose flow rate is determined by the steady - state solution of darcy s law . tracer particles passively follow the fluid motion and advance with the average flow velocity @xmath6 . the width of the tracer density profile spreads as @xmath7 due to hydrodynamic dispersion ( fig . [ fig1 ] ) . invader particles first encounter clean collector sites . because each such encounter leads to deposition with probability @xmath0 , the survival probability of the particles in this leading invaded region decreases exponentially with downstream position as illustrated in fig . as particles advance and get trapped , the pore capacity decreases and subsequent particles are more likely to survive , giving rise to an advancing invasion front with a velocity @xmath8 . in principle , the propagation velocity and shape of the front are functions of time . however , at long times these features approach steady - state values . feature is that the trailing edge of the capacity profile decays as a power of the distance @xmath9 ( @xmath10 ) from the invasion front whose location is defined , _ e.g. _ , as the position where @xmath11 . for large @xmath9 , any reasonable definition for the front location can be used . the existence of different propagation velocities , @xmath6 for the pure fluid and @xmath8 for the contaminant , leads to purification of the liquid . the filter can be used until the invaded region reaches the outlet end . for a filter of length @xmath12 , the breakthrough time will be of the order of @xmath13 , so the amount of throughput will be approximately proportional to @xmath14 . 0.25 cm .6 cm as a preliminary , we study infiltration in a one - dimensional chain of identical pores at @xmath15 . first we consider the case where each pore can accommodate only one particle and then we generalize to multiple capacity pores . we choose a time unit such that one particle is injected at each discrete time step . multiple particle injection leads to a different particle density and will affect only the overall scale factor and not change qualitative features of the system . the carrier fluid advances by one pore distance at each time step ; that is , its velocity is unity . when particle trapping occurs in a pore at @xmath16 , the capacity @xmath17 changes permanently from 1 to 0 . at time @xmath3 , a particle at pore @xmath16 gets trapped in that pore with probability @xmath0 if @xmath18 . a particle advances to the next pore in one time step with probability 1 if @xmath19 . based on these elemental steps , we introduce the following two probability densities : * @xmath20 : the probability that a freely - moving particle is in pore @xmath16 at time @xmath3 . * @xmath21 : the probability that the pore at site @xmath16 is unoccupied ; that is @xmath18 . the corresponding master equations for @xmath20 and @xmath21 are @xmath22 where we drop the argument @xmath3 on the right hand side for simplicity . unless there is a possibility for confusion , we will not write the argument @xmath3 in related formulae . since a particle advances to the next pore in one time step , @xmath23 depends on @xmath24 . the term @xmath25 in eq . ( [ 1dmep ] ) is the probability that the particle at @xmath26 does not get trapped by an unoccupied pore also at @xmath26 . similarly , the term @xmath27 in eq . ( [ 1dmeq ] ) is the probability that pore @xmath16 does not trap a free particle at time @xmath3 . the initial and boundary conditions for these equations are : @xmath28 if the trapping probability @xmath0 is small , a particle can advance many pores without being trapped , so that @xmath20 and @xmath21 vary slowly in space and time and a continuum approximation can be applied . letting @xmath29 and @xmath30 , eqs . ( [ 1dmep ] ) and ( [ 1dmeq ] ) become , to lowest order , @xmath31 where @xmath32 , and @xmath33 is a redefinition of the trapping probability in units of the infinitesimal time increment . in a co - moving reference frame , @xmath34 , with @xmath35 the invasion front propagation velocity shown in fig . [ fig1 ] ( which is yet to be determined ) , eqs . ( [ pq ] ) become @xmath36 let us first examine the steady state solution of eqs . ( [ 1dpwave ] ) and ( [ 1dqwave ] ) . setting the time derivatives to zero , subtracting eq . ( [ 1dqwave ] ) from eq . ( [ 1dpwave ] ) , and integrating with respect to @xmath37 gives @xmath38 the integration constant can be determined by applying eqs . ( [ 1dic ] ) in the co - moving frame . as @xmath39 , @xmath40 , @xmath41 , and as @xmath42 , @xmath43 , @xmath44 . these immediately give @xmath45 . note that @xmath35 is determined entirely from the boundary condition ( and @xmath6 ) in the co - moving frame , and not from the interaction strength @xmath0 . this feature continues to hold for all the models in this paper . using eq . ( [ 1dpqsol ] ) with @xmath46 , eqs . ( [ 1dpwave ] ) and ( [ 1dqwave ] ) can be solved to give @xmath47 ( fig . [ fig2 ] ) . thus @xmath48 is the characteristic width @xmath49 of the profile . notice also that the profiles of @xmath50 and @xmath51 are symmetric about their intersection . we verified both the dependence of the width on @xmath0 and the profile shape predicted by eq . ( [ 1dpq ] ) by numerical integration of the master equations ( [ 1dmep ] ) and ( [ 1dmeq ] ) . .1 cm one subtle point is the rate of approach to the steady state . first , we find that the asymptotic propagation velocity @xmath46 is reached before the asymptotic profile is established . this arises because @xmath35 is determined by the boundary conditions , and not by interaction details . we also verified this feature numerically . adopted this asymptotic velocity , eqs . ( [ 1dpwave ] ) and ( [ 1dqwave ] ) are then symmetric in @xmath50 and @xmath51 . in fact , the system is identical to two - species annihilation , @xmath52 , where each species is ballistically injected from opposite sides with velocity @xmath53 for the @xmath54s and @xmath55 for the @xmath56s . thus most of the the time variation in fig . [ fig2 ] occurs in the reactive region of width @xmath49 , where @xmath57 at @xmath5 . integrating eq . ( [ 1dpwave ] ) from @xmath58 to @xmath59 gives , @xmath60 the integral on the left hand side is the area under the curve @xmath61 between @xmath62 and @xmath63 , whose time dependence mainly comes from the change in @xmath49 . on the right hand side , the integral is approximately proportional to @xmath49 , since @xmath64 is significantly different from zero only in the reactive region . we can then rewrite eq . ( [ 1dpint ] ) as @xmath65 where @xmath66 and @xmath67 are constants . integrating eq . ( [ 1dpint1 ] ) and applying the condition @xmath68 gives an exponential decay to the steady state @xmath69 . it is worth emphasizing that the symmetry between the invader and defender is generally responsible for the relation @xmath46 . at the inlet , invaders are injected with velocity @xmath6 , and the invasion front advances with velocity @xmath35 , with one invader particle annihilating with one defender site . in the reference frame moving with velocity @xmath6 , the situation is reversed . the invaders are at rest and defenders are injected with velocity @xmath6 from the opposite direction . therefore the invasion front advances with velocity @xmath70 . since these two reference frames describe the system in the same way , the front velocities should be the same ; that is , @xmath71 , or @xmath46 . now we consider the case where each pore can trap @xmath72 particles , that is , the initial pore capacity is @xmath73 . we again follow the previous rules of injecting a single particle and advancing a particle by one pore ( @xmath74 ) at each time step . multiple particle injection or different injection intervals again simply changes the overall concentration and time scale . in a multiple capacity pore , the probability of encountering an open trap in a pore needs be considered , in addition to the trapping probability @xmath0 upon encounter with an open trap . generally , the encounter probability decreases as more particles get trapped , since the inner pore surface area available for trapping shrinks . when fluid mixing within a pore is weak , a particle can encounter only one trap , either open or occupied . then the encounter probability is approximately proportional to the fraction of the open surface area . on the other hand , if the mixing is perfect , a particle encounters all the available traps in a pore before exiting . for practically relevant situations , pores are sufficiently short so that a particle in a pore follows streamlines without transverse diffusive mixing @xcite . in what follows , we consider this limit of weak mixing . for a pore with @xmath75 out of @xmath72 traps available , the encounter probability is @xmath76 , and the overall trapping probability of this pore is @xmath77 . in writing this expression , we ignore the possibility that a particle far from the pore wall does not encounter any traps . in sec . [ discussion ] , we argue that this volumetric effect does not change the basic behavior of infiltration . to describe the evolution of the system , we use the same single - particle probability density @xmath20 as in the single - capacity pore system , but modify the probability density for the capacity as follows : * @xmath78 : the probability that a pore at position @xmath16 contains @xmath75 open traps . this is the same as the probability that @xmath79 , for @xmath80 . following similar reasoning as that applied to deduce eqs . ( [ 1dmep ] ) and ( [ 1dmeq ] ) , the master equations for @xmath20 and @xmath78 are @xmath81\\ & & \nonumber\\ \label{mulmeq } q_k^n(t+1 ) & = & q_k^n(1-p_k{t}_n ) + p_k q_k^{n+1}{t}_{n+1}.\end{aligned}\ ] ] in eq . ( [ mulmep ] ) , @xmath82 accounts for the case that the pore @xmath26 has zero capacity . other terms in eq . ( [ mulmep ] ) correspond to cases when the capacity is different from 0 , with @xmath83 the survival probability for each case . in eq . ( [ mulmeq ] ) @xmath84 is the probability that the pore with capacity @xmath75 does not trap a particle , and @xmath85 is the probability that the capacity decreases from @xmath86 to @xmath75 by a particle trapping event . hence the last term is absent when @xmath87 . we simplify eqs . ( [ mulmep ] ) and ( [ mulmeq ] ) by introducing the _ average capacity _ of a pore at position @xmath16 , @xmath88 this gives the average number of sites still available for trapping in the pore . now by multiplying eq . ( [ mulmeq ] ) by @xmath75 , summing from 1 to @xmath72 , and using @xmath89 , we obtain @xmath90 these are identical in form to eqs . ( [ 1dmep ] ) and ( [ 1dmeq ] ) , so the same steady state analysis applies . we transform to a co - moving frame and take the continuum approximation to reduce the rate equations to eqs . ( [ 1dpwave ] ) and ( [ 1dqwave ] ) with @xmath91 and @xmath92 . the boundary conditions are also the same as in the case of single - capacity pores , except @xmath93 . combining these results give @xmath94 notice that for @xmath95 , eq . ( [ mulvw ] ) reduces to the single capacity case , while for @xmath96 , @xmath97 . this means that there is no steady state for the case of infinite capacity pores . we can generalize the symmetry argument given in the single capacity case to find the propagation velocity in eq . ( [ mulvw ] ) . at the input , the flux of invaders moving with the carrier fluid is equal to @xmath98 . similarly , in the reference frame moving with velocity @xmath6 , the flux of defenders is @xmath99 , while the invaders are at rest . because one invader annihilates with one defender , the two particles are kinetically indistinguishable . therefore , if a particle flux of @xmath98 results in a front moving with velocity @xmath35 , the front velocity produced by a flux of @xmath99 should be @xmath100 , which , in turn , equals @xmath70 in the moving reference frame . by this equivalence , eq . ( [ mulvw ] ) immediately follows . in the limit of perfect mixing , a particle encounters all traps in the pore . the overall trapping probability with @xmath75 open traps is then @xmath101 . in the limit of small @xmath0 , @xmath102 , thus the analysis is exactly the same as in the poor mixing case except without the factor @xmath103 in @xmath104 . the propagation velocity of the front is the same as in eq . ( [ mulvw ] ) , since this velocity is independent of trapping mechanism , while the width varies as @xmath105 . notice that @xmath49 is a decreasing function of @xmath72 . this arises because a particle must survive all the traps in a pore before advancing to the next pore . finally , for a mixing mechanism which is intermediate between the two limits of perfect and poor mixing , the propagation velocity will be @xmath106 , while the width of the front will lie between the limiting values of @xmath107 and @xmath108 . we now study the bubble model as a logical next step towards understanding infiltration in porous media . the bubble model was introduced to account for the breaking of fibers @xcite , extremal voltages in resistor networks @xcite , and later to filtration kinetics @xcite . the bubble model consists of @xmath12 `` bubbles '' in series , each of which is a parallel bundle of @xmath49 tubes , with each tube representing a pore ( fig . [ fig4 ] ) . a bubble can be viewed as a single layer of parallel bonds in a lattice with all the ends `` shorted '' . this model has multiple paths , as in real porous media , and is sufficiently simple to be amenable to analytic study . a useful feature of the bubble model is that for straining dominated filtration , this model predicts similar behavior to that of lattice networks @xcite . .5 cm .4 cm we choose the tube radii in the bubble model from the hertz distribution @xmath109 where @xmath110 is the characteristic pore radius . this form is often seen in experimental pore size measurements @xcite and has been used for modeling the pore size distribution in filters @xcite . for simplicity we assume identical tube lengths and measure downstream distance in units of the tube length , which is set equal to 1 . we also assume that the flow rate in a tube of radius @xmath111 is proportional to @xmath112 , where @xmath113 is the pressure gradient along the tube and @xmath114 depends on the nature of the flow , with @xmath115 corresponding to poiseuille flow and @xmath116 to euler flow . perfect mixing is assumed at each node . a particle chooses a tube in the next downstream bubble according to _ flow induced probability _ @xmath117 , @xmath118 in which the probability of choosing an outgoing tube of radius @xmath111 is proportional to the flow rate into the tube , @xmath119 @xcite . since tubes of different flow velocities give the dominant mechanism for dispersion , the radial dependence of the local flow velocity in a tube ( taylor dispersion ) is ignored . thus we assume that a particle moves with the average flow velocity @xmath120 along the tube . we now investigate the hydrodynamic dispersion of passive brownian particles which are carried by the background fluid in the bubble model , in the absence of any trapping . this will provide the concepts and tools necessary to understand infiltration in the bubble model . in the large @xmath49 limit , each bubble is nearly identical , and we can regard the particle motion as a directed random walk in which the average residence time @xmath121 in bubble @xmath16 ( @xmath15 ) is a random variable whose distribution @xmath122 is related to the flow induced entrance probability @xmath117 and the radius distribution @xmath123 . this random walk description of the continuous particle motion introduces an additional stochasticity into the system . however , we will show below that this only modifies the hydrodynamic dispersion coefficient by an overall multiplicative factor . the master equations for @xmath20 , the probability that there is a particle in the @xmath124 bubble at time @xmath3 , are @xmath125 here @xmath126 is the initial particle number concentration , @xmath127 is the ( constant ) flow rate , and the initial condition is @xmath128 for all @xmath16 . since the flow rate does not change in infiltration , constant pressure drop and constant flow rate conditions are equivalent . the particle transport properties can be obtained in terms of the residence time distribution @xmath122 , namely , the probability that a particle spends a time @xmath129 in a bubble . this residence time distribution is related to microscopic distributions by @xmath130 where @xmath131 is the dirac delta function . since the flow rate into a tube of radius @xmath111 is @xmath132 , the average flow velocity @xmath133 . using this together with eq . ( [ hertz ] ) for @xmath123 , we obtain the first two moments of @xmath129 @xmath134 where @xmath135 is the average tube volume ( recall that the tube length is fixed to be 1 ) and @xmath136 is the gamma function . we solve eq . ( [ bubbleme ] ) in the appendix by the laplace transform technique . from this solution , the average propagation velocity and the width of the front are @xmath137\right\}^{1\over2}\equiv ( d_\parallel t)^{1\over2}.\end{aligned}\ ] ] thus we see that the dispersion coefficient is proportional to the average flow velocity @xmath138 . when @xmath116 ( euler flow ) , the flow velocities in all the tubes are identical and there should be no dispersion . however , eq . ( [ hydrow ] ) gives a nonzero dispersion coefficient . as mentioned above , this arises from the stochasticity of the random walk picture for the particle motion . for the practically relevant case of @xmath139 , the effect of this stochasticity is only to change the dispersion coefficient by a factor of order unity . to describe infiltration in the bubble model , we need to specify the particle motion , the tube capacities , and particle trapping in a tube . for the particle motion we again assume that a particle chooses a tube according to flow induced probability and then advances with the average flow velocity @xmath120 of this tube . the capacity of a tube is proportional to its inner surface area , which is proportional to the tube radius , since all tubes have the same length . last , the overall trapping probability of a tube is equal to the microscopic trapping probability @xmath0 multiplied by the fraction of open traps in a tube ( sec . [ 1dmulti ] ) . to simulate this process efficiently we propagate the probability distribution function ( pdf ) of the suspended particles rather than simulating the motion of individual particles@xcite . the pdf propagation therefore provides the exact distribution of particle positions and tube capacities for a single realization of tube radii . conceptually , the pdf algorithm is equivalent to an exact integration of the master equations . to implement the pdf propagation , we define * @xmath140 : the probability that there is a particle at the _ entrance _ of tube @xmath141 in bubble @xmath16 ( @xmath142 , @xmath143 ) . * @xmath144 : the capacity of tube @xmath141 in bubble @xmath16 . since particles generally have different velocities , their positions could be anywhere within a tube . we simplify this by forcing particles to _ always _ be at the tube entrance by adjusting the time unit and the pdf propagation so that the _ average _ particle position is at the correct location along the tube , as illustrated in fig . [ fig5 ] . to construct the particle motion , let us temporarily disregard particle trapping . we set the time increment to be @xmath145 , where @xmath146 is the maximum flow velocity among all tubes . in a time @xmath147 , a particle at the entrance of the fastest tube should traverse the entire tube length which is equal to 1 . we then let a particle in a slower tube , with velocity @xmath148 , travel a distance @xmath149 with probability @xmath150 , or remain fixed with probability @xmath151 . one can regard @xmath152 as a normalized flow velocity . by construction , such a particle travels the correct average distance in time @xmath147 , @xmath153 . let us now recast this random walk into a probability propagation algorithm . consider an element of the pdf which is at the junction before the @xmath154 bubble . before any particle motion occurs , we split this probability element among the downstream bonds in this bubble according to the flow induced probability at the tube entrance . we can view the probability element as advancing infinitesimally into each bond , as indicated on the left side of fig . once this initial tube assignment is made , the probability element remains within its assigned tube until it reaches the next junction . now consider the motion of a probability element which has just entered a particular bond . after a time @xmath147 , a fraction @xmath152 of the pdf is advanced to the next bubble , while a fraction @xmath151 remains fixed at the entrance to bond @xmath141 . .4 cm due to the filtration , a fraction of the flowing pdf becomes trapped in tube @xmath141 in the @xmath124 bubble at a rate which is proportional to the tube capacity @xmath144 . the overall trapping probability of this tube is therefore @xmath155 . after trapping has occurred , the tube capacity is decremented according to the following prescription . when one unit of pdf ( equivalent to one particle ) gets trapped , we define the bond capacity to be decreased by @xmath156 . therefore @xmath156 is just the surface area covered by one particle . correspondingly , @xmath157 equals the number of particles the tube can accommodate . our algorithm for propagating an element of probability at the entrance to bond @xmath141 in the @xmath124 bubble over a time @xmath147 therefore consists of the following steps ( fig . [ fig5 ] ) : * fraction of pdf remaining at the start : @xmath158 . * fraction trapped in tube @xmath141 : @xmath159 . * fraction advancing to the next junction : + @xmath160 * capacity change of the tube by trapping : + @xmath161 $ ] . the rate equations which account for these steps are : @xmath162 the first term on the right hand side of eq . ( [ bubblepdf ] ) is the fraction of probability that does not move , and the second term is contribution from elements of probability which has moved from the previous site . the flow induced probability @xmath163 in eq . ( [ bubblepdf ] ) accounts for the fraction of pdf which enters into tube @xmath141 . to test this approach , we set @xmath164 ( no trapping ) in the above rate equations and simulate the pdf propagation . by this method , we find a traveling front whose basic properties coincide with the hydrodynamic dispersion results given by eqs . ( [ hydrov ] ) and ( [ hydrow ] ) . it is also worth mentioning that our pdf algorithm can be generalized to allow for hopping a distance which is a fraction of the tube length . in this manner one can account for different longitudinal flow velocities at different radial positions within a tube ( taylor dispersion @xcite ) . in the limit of an infinitesimal hopping distance , continuous particle motion is reproduced by the pdf algorithm . unfortunately , the gain in having a more accurate description of the motion is offset by the complexity of the algorithm and large increase in the computation time . to obtain the average invasion front profile over the tubes in each bubble , we define the bubble - average quantities @xmath165 and @xmath166 . we first derive the invasion front velocity via the symmetry argument of sec . [ 1dmulti ] . a rigorous derivation of the front velocity from the master equations for @xmath167 and @xmath168 is given in @xcite . the carrier fluid is moving with velocity @xmath6 ( eq . ( [ hydrov ] ) ) and the input flux of invaders per tube is equal to @xmath169 , since @xmath170 is equal to the number of invader particles per tube volume in the input fluid . on the other hand , in a reference frame moving with velocity @xmath6 , the input flux of defenders is equal to @xmath171 where @xmath172 is the average initial number of invaders a tube can accommodate . following the argument in sec . [ 1dmulti ] , we find the front velocities @xmath35 and @xmath70 in the two reference frames are related by @xmath173 , yielding @xmath174 good filter performance means that the breakthrough time is long or , equivalently , that the propagation velocity is slow . ( [ bubblev ] ) implies that the propagation velocity can be made small by increasing the capacity of a pore , or by decreasing either the filter grain size or the input particle concentration . notice that neither the reaction strength @xmath0 nor the nature of the flow ( through @xmath175 ) affect this propagation velocity . we now study the asymptotic density profiles . instead of working directly with the averaged quantities @xmath167 and @xmath168 , we first focus on the behavior of a single tube of radius @xmath111 , since tubes with the same radius in a bubble have identical time dependence . the asymptotic profiles can be obtained after averaging over the distribution of tube radii . therefore , we label tubes according to their radii instead of the index @xmath141 . we denote @xmath176 and @xmath177 as the pdf and capacity of a tube of radius @xmath111 in the @xmath124 bubble . let us first focus on the pdf profile in the invaded region . here , traps are mostly unoccupied , so that the tube capacity @xmath177 is approximately equal to its initial value @xmath178 , which is proportional to the tube radius and we set it equal to @xmath111 . the arbitrariness in the unit of capacity can be controlled by the magnitude of the parameter @xmath156 . then eq . ( [ bubblepdf ] ) becomes , @xmath179 where the integration over @xmath180 replaces the summation over the tube index , the flow induced probability @xmath117 and normalized velocity @xmath181 are independent of the downstream position @xmath16 because these only depend on the radius of a tube . the equation for @xmath182 can be obtained by multiplying eq . ( [ attack ] ) by @xmath123 and integrating over @xmath111 . as in sec . [ 1d ] , we take @xmath183 and consider the continuum limit . if we redefine the length of the bubble from 1 to @xmath184 , @xmath147 becomes @xmath185 . integrating eq . ( [ attack ] ) and expanding in @xmath147 and @xmath184 yields @xmath186\ ] ] here , we use @xmath187 . dividing eq . ( [ attack0 ] ) by @xmath147 changes @xmath181 back to @xmath120 . after redefining @xmath188 as before , we obtain @xmath189.\ ] ] in the steady state co - moving frame , eq . ( [ attack1 ] ) becomes @xmath190.\ ] ] since only a small number of particles have entered the invaded region , the density of moving particles is approximately proportional to @xmath119 , and we introduce the ansatz @xmath191 to factorize the pdf . in order to calculate the dominant contribution from the integral in eq . ( [ attack2 ] ) , we substitute @xmath192 , where @xmath193 is the deviation from the average carrier fluid velocity @xmath6 . since @xmath193 has zero mean , the dominant contribution to the integral over @xmath120 in eq . ( [ attack2 ] ) comes from the constant part @xmath6 . using these approximations in eq . ( [ attack2 ] ) , and using @xmath194 , we find @xmath195 since @xmath8 , we find @xmath196 $ ] . hence the profile of free particles in the invaded region @xmath197 decays exponentially in @xmath37 , with a characteristic decay length which has the same @xmath198 dependence as in the 1d model . let us now turn to the analysis of the capacity profile in the tail region . in terms of @xmath199 and @xmath200 , eq . ( [ bubblecap ] ) becomes @xmath201 since there is negligible trapping in the tail region , the particle motion follows that of the carrier fluid . thus @xmath202 , where @xmath203 is the tube volume , and the flow velocity is @xmath204 . substituting these in eq . ( [ tail ] ) and transforming into the co - moving frame gives @xmath205 where @xmath206 denotes the strength of the particle trapping reaction . we now integrate eq . ( [ tail1 ] ) from @xmath58 to @xmath207 and use the boundary condition @xmath208 to obtain @xmath209 where we drop the subscript of @xmath210 . finally , the average bond capacity as a function of position with respect to the front , @xmath211 , is @xmath212 for large @xmath9 , the integral is dominated by the smallest tubes and the initial distribution of tube radii is irrelevant in the tail region . hence the factor @xmath213 in the exponential can be ignored . performing the resulting integration gives @xmath214 where the last relation serves to define the _ profile exponent _ @xmath215 . this is one of our primary results . correspondingly , the pdf in the tail region will approach its initial value with the same power law . the existence of the power law tail in the capacity profile stems from the fact that the flow rate is not affected by trapping . thus when large pores are `` used up '' , the fluid still predominantly flows through these pores , leading to a substantial unused capacity in the smaller tubes . it is these unused smaller tubes which contribute substantially to the capacity profile in the tail region . this mechanism is quite general and only depends weakly on the form of the radius distribution . for example , for a uniform distribution in the range @xmath216 , we obtain @xmath217 . however , if there is a finite lower cutoff in the radius distribution , the pdf will have an asymptotic exponential tail . it is interesting to note that the density profile has different dependence on @xmath0 in the invaded and tail regions . from eq . ( [ tail4 ] ) , the density profile contains an overall factor @xmath218 . thus @xmath0 typically does not appear as an overall scale factor of the entire profile , as in the invaded region . however , for the practically relevant case of @xmath115 , the exponent in eq . ( [ tail4 ] ) is equal to 1 , and @xmath0 becomes the overall scale factor of the profile . in our numerical simulations , we set the input particle flux per tube @xmath219 equal to 1 , which means that @xmath49 units of pdf are injected into the system at every time step . this can be achieved by choosing @xmath220 and @xmath221 , which also makes @xmath222 ( eqs . ( [ tau1 ] ) and ( [ hydrov ] ) ) . .5 cm .4 cm we applied the pdf propagation of eqs . ( [ bubblepdf ] ) and ( [ bubblecap ] ) to a system of size @xmath223 . due to the exact nature of the pdf algorithm , a single realization provides good quality data for @xmath224 tubes . a system length of @xmath225 is sufficiently long to give the continuum functional form of the profiles . all the data shown below are results of single realization of tubes including the network simulation in the next section . the simulation is stopped before the front exits the system . _ density profiles_. fig . [ fig6 ] shows typical particle and tube capacity profiles . there are strong bubble - to - bubble fluctuations and some type of smoothing procedure is necessary . we use the savitzky - golay smoothing technique , which approximates successive windows of data points to a @xmath226 order polynomial ( solid lines in the figure ) @xcite . this technique is superior to local averaging because savitzky - golay smoothing can faithfully follow rapid changes in the profile , as can be seen in fig . this smoothing is also useful in estimating the exponents . since logarithm of the profile in the tail region amplifies the fluctuation in nonlinear way , slopes of the raw data in fig . [ fig_tail1 ] are larger than those from the smoothed data , and differs from the predicted value of the profile exponent @xmath215 . .1 cm .4 cm _ front velocity_. fig . [ fig_vel](a ) shows the front position , defined as the point where @xmath167 is half of its saturation value , versus time . notice that a constant front propagation velocity sets in almost immediately . with @xmath220 and @xmath221 , ( [ bubblev ] ) gives @xmath227 the slopes in fig . [ fig_vel](a ) agree well with eq . ( [ vel ] ) . notice that the propagation velocity does not depend on the reaction strength @xmath0 nor the exponent @xmath114 in the radius dependence of the velocity . .5 cm _ tail profile_. fig . [ fig_tail1](a ) shows the tube capacity profile @xmath211 in the tail region as a function of the distance @xmath9 ( @xmath10 ) from the front on a double logarithmic scale . the plot becomes straight for large @xmath9 and the slope in this region corresponds to the exponent @xmath228 predicted by eq . ( [ tail4 ] ) . for the uniform distribution on @xmath229 , we predicted the profile exponent to be @xmath217 . for @xmath115 , the exponent value of 2/3 agrees well with our simulations ( fig . [ fig9 ] ) . however , for a radius distribution with a lower size cutoff , we expect an exponential density profile ( inset to fig . [ fig9 ] ) . .1cm.4 cm .5 cm .4 cm as we also discussed in sec . [ bubbleanal ] , the amplitude of the density profile in the tail region typically has a power - law dependence on @xmath0 . for a hertz distribution of particle radii , this amplitude should be proportional to @xmath218 according to eq . ( [ tail4 ] ) . thus fig . [ fig_tail2](a ) shows @xmath230 versus @xmath231 for @xmath115 and @xmath232 . values of the abscissa should be proportional to @xmath198 , which is indeed the case . _ invader profile_. fig . [ fig_invade](a ) is a log - normal plot of the invader profile @xmath197 ( raw data ) versus @xmath37 . the slopes of the two data sets are in excellent agreement with the predicted value @xmath233 from eq . ( [ attack3 ] ) . in the invaded region , since there are few particles present , the corresponding pdf monotonically decreases and its fluctuation is significantly smaller compared to the tail region . -.2cm.4 cm .2 cm .4 cm we now consider infiltration on a square lattice network of tubes . here , local mixing at tube junctions occurs as opposed to the mean - field - like mixing in the bubble model . nevertheless , many of our predictions from the bubble model continue to be valid for the lattice network . for example , we expect that the propagation velocity given by eq . ( [ bubblev ] ) will continue to hold in the lattice network because it is determined _ only _ by the boundary conditions of the particle flux and initial filter capacity . we also find numerically that the network model has both the same exponential invader profile and the power law capacity profile as in the bubble model , although the values of the amplitudes and decay exponents are different . overall , it appears that the bubble model provides an excellent account of the numerical results from the lattice network . we study a square lattice of size @xmath234 which is tilted at @xmath235 . a periodic boundary condition is imposed in the transverse direction . the tube radii are drawn from the hertz distribution of eq . ( [ hertz ] ) . notice that the bubble model would arise from this system by merging together all sites at the same longitudinal position . the overall flow rate @xmath127 is set to @xmath236 and the particle density to @xmath220 , just as in the bubble model . for a given set of tube radii , the flow field is calculated by using the conjugate gradient method@xcite to solve the set of linear algebraic equations for fluid conservation at each node . the tolerance of the computation is set so that the measured average pdf and the tube capacity in the @xmath124 layer are accurate to within 0.01% . after the flow field is solved , we use the same pdf algorithm as in the bubble model to track the motion of the suspended particles . the new features due to the lattice nature of the network are that tubes are only locally connected and that the local flow direction is not always downstream . to facilitate comparisons , all our numerical results for the bubble model and the square lattice are presented side - by - side . [ fig_vel ] shows the front position versus time for both the bubble model and the square lattice . the square lattice results are in excellent agreement with the bubble model prediction for the front velocity , eq . ( [ vel ] ) . similarly , the tube capacity profile exhibits a power law tail ( fig . [ fig_tail1](b ) ) . however , the dependence of the decay exponent @xmath215 on @xmath114 is much weaker compared to the bubble model . as @xmath114 decreases from 4 to 2 , @xmath215 increases slowly from 0.95 to 1.28 . to isolate the dependence of the capacity profile on the trapping probability @xmath0 , fig . [ fig_tail2](b ) shows @xmath230 versus @xmath231 in the tail region . here , the values of @xmath215 used are obtained from fig . [ fig_tail1](b ) . unlike the bubble model , the overall amplitude has a relatively stronger dependence on @xmath114 . however , the dependence on @xmath198 still holds even in the network case . it seems that in the square lattice network , @xmath114 has more effect on the amplitude of the tail than on the decay exponent . lastly , fig . [ fig_invade](b ) shows the density profile of invaders in the invaded region . the slopes of these particles are almost identical to those of the bubble model . currently we do not have clear explanation of this fact . it seems that the network geometry does not affect the profile . in fact , the characteristic decay length , @xmath237 , of the profiles in fig . [ fig_invade ] is the same as that of the corresponding 1d model in sec . [ 1dmulti ] where @xmath170 invaders are injected with velocity @xmath6 into the chain of defenders of capacity @xmath72 . in this paper , we studied infiltration , in which suspended particles are removed from a carrier fluid as the suspension passes through a porous medium . the trapping mechanism has a built - in saturation so that once all available trapping sites are used up , subsequent particles can pass through the medium freely . the particles are assumed to be sufficiently small that their trapping does not change the flow rate . the basic dynamical properties of this infiltration process are the density profile of the invader particles and the capacity profile of the remaining active pores . when the invader profile reaches the end of the system , the output concentration of particles quickly increases to a saturation level and the filter should be discarded . thus the features of this profile are important to understand the operating characteristics of infiltration . we have developed a series of discrete network models to describe the basic characteristics of infiltration , starting with a one - dimensional model and building up to the bubble model , which is a series array of parallel , multiple - capacity tubes . the advantage of these quasi - one - dimensional models is that they remain relatively simple , even after incorporating local spatial heterogeneity . the bubble model , in particular , appears to capture many of the quantitative features that we observed in numerical simulations of infiltration on a square lattice tube network . our modeling is also flexible , so that variations can be easily implemented for case - specific situations . our main qualitative result is that basic dynamical features of the system , including the value of the front propagation velocity , the exponential profile of flowing particles in the invaded region , and the power law capacity profile of pores in the tail region , are relatively insensitive to microscopic details of the model . we have also identified the basic parameters which do affect quantitative features of the profiles . it is useful to summarize these results and to compare with experimental data , as well as with predictions from previous studies based on the reaction - diffusion equation approach @xcite . .2 cm .4 cm _ invader concentration at the output_. in typical experiments , the invader concentration at the output is measured as a function of time . a slower propagation velocity shifts this output concentration curve to a later time , as indicated in fig . [ fig_discuss](a ) . typically , the time unit is normalized by the time for passive particles to pass through the system @xcite . hence , the amount of the time shift is determined by the ratio @xmath238 rather than by @xmath35 itself . a nice set of infiltration experiments , as well as an accompanying numerical study of the reaction - diffusion equation were performed in @xcite . in these experiments , contaminant solutions with different values of the invader particle diameter @xmath239 , but with fixed mass concentration were used . this makes the corresponding number density @xmath126 of invaders in each solution proportional to @xmath240 . also , since the cross - sectional area of each particle is proportional to @xmath241 , the average initial tube capacity @xmath72 varies as @xmath242 . from eq . ( [ bubblev ] ) , we then have @xmath243 . thus the output concentration curve shifts to a later time for a solution with a larger value of @xmath239 , which is consistent with the experiment and numerical predictions in @xcite . in another set of experiments , different electrolyte concentrations of the carrier fluid were used . this mainly affects the trapping rate @xmath0 . for a larger trapping rate , the width of the output concentration curve , namely , the time range over which the output concentration changes from zero to its saturation value becomes narrower . however , there is no shift in the breakthrough time because the propagation velocity is independent of @xmath0 . these two features are illustrated in fig . [ fig_discuss](b ) . this behavior again qualitatively agrees with the experiment in @xcite . it would also be interesting to study the effect of different filter grain sizes . in our model this would be accomplished by changing the characteristic parameter @xmath244 of the pore size distribution . in turn , this affects the invader propagation velocity , and the output concentration curve will shift in time accordingly . however , since the microscopic parameters we use in our modeling may be coupled with each other in experimental situations , different sizes of filter grains or invader particles may also affect , _ e. g. _ , the reaction strength @xmath0 . we can incorporate such a coupling effect by extending our model to deal with these effects explicitly . for example , we can adapt the microscopic models of particle trapping on a single sphere or on a plane @xcite , to the tube geometry . from such an approach , we can express the reaction strength @xmath0 as a function of the invader or defender diameter . _ tail of the output concentration curve_. a slowly decaying tail in the deviation of the output concentration from its asymptotic value is generally observed in experiments @xcite . this observation is in contrast to the empirical approaches , such as that given in @xcite , which gives an exponential profile for the whole time range . their prediction agrees with experimental observations at early times but then deviates at later times , implying that the output profile at later times is not exponential . a closely related approach , based on the study of a reaction - diffusion equation , is presented in @xcite , along with experiments which measure the output concentration . unlike @xcite , here the adsorption rate depends on the local concentration of contaminants and thus is spatially inhomogeneous . a crossover from a rapid increase to a slowly decaying tail of the output concentration was numerically predicted . however , the functional forms of these two regimes in particular , whether they are exponential or power law in time were not investigated quantitatively . however , the data presented in this work seem consistent with a slower than exponential decay of the density profile . in @xcite , an exponential output concentration profile , @xmath245 , is assumed from the outset , where @xmath2 is the downstream distance and @xmath246 is the experimentally measured filter coefficient . the corresponding experimental data show that @xmath246 is constant at early times , and then sharply decreases at later times . thus the initial stage of the experiment is consistent with an exponential profile , but later the profile decays more slowly . in @xcite , this is attributed to a `` blocking effect '' in which previously - deposited particles can block the further deposition of particles onto nearby available trapping sites . _ probability of encountering an open trap_. as a last remark , let us examine the assumption that the probability of encountering an open trap is proportional to the fraction of open traps in a pore ( sec . [ 1dmulti ] ) . suppose instead that one takes into account the volumetric effect that particles far from the surface of the pore do not have chance to encounter a trap . then the fraction of particles in contact with the inner surface of a tube is proportional to @xmath247 , namely , the ratio between the surface area and the volume of the tube . this would lead to the interaction terms involving @xmath248 in eqs . ( [ bubblepdf ] ) and ( [ bubblecap ] ) being multiplied by another factor of @xmath247 . however , this modification does not affect the propagation velocity of the front , nor the power law feature of the tail . only the decay exponent changes through the steps of eqs . ( [ tail])-([tail3 ] ) with an additional factor of @xmath247 . our results can also provide practical guidelines for improving the design of a filter in two aspects , namely , the breakthrough time and the amount of filter material used before the breakthrough . a longer breakthrough time can be achieved by having a smaller filter grain size , a lower input concentration , or a larger pore capacity . while these trends may seem intuitively clear , we can quantitatively estimate the increase in the breakthrough time through the expression for the propagation velocity , eq . ( [ bubblev ] ) . when breakthrough occurs , the amount of unused filter material is determined by the shape of the tail in the density profile of the defenders . according to eq . ( [ tail4 ] ) , the amplitude of this tail is proportional to @xmath218 . from this , we can quantitatively estimate the amount of filter material left unused at the breakthrough time as a function of the reaction strength . we thank dr . jysoo lee for helpful discussions about flow field calculations . we are also grateful to grants aro daad19 - 99 - 1 - 0173 and nsf dmr9978902 for financial support . 99 c. tien and a. c. payatakes , aiche j. * 25 * , 737 ( 1979 ) . r. f. probstein , _ physicochemical hydrodynamics _ , 2nd ed . ( john wiley & sons , 1994 ) . a. o. imdakm and m. sahimi , phys . rev . a * 36 * , 5304 ( 1987 ) . l. lhrmann , u. noseck , and c. tix , water res . res . * 34 * , 421 ( 1998 ) . l. bloomfield , scientific american , p152 ( december , 1999 ) . d. putnam and m. burns , chem . sci . * 52 * , 93 ( 1997 ) . h. brenner and d. a. edwards , _ macrotransport processes _ ( butterworth - heinemann , 1993 ) . r. rajagopalan and c. tien , aiche j. * 22 * , 523 ( 1976 ) . j. lee and j. koplik , phys . e * 54 * , 4011 ( 1996 ) . c. ghidaglia , l. de arcangelis , j. hinch , and e. guazzelli , phys . e * 53 * , r3028 ( 1996 ) . s. datta and s. redner , phys . e * 58 * , r1203 ( 1998 ) ; s. datta and s. redner , int . c * 9 * 1535 ( 1998 ) . a. o. imdakm and m. sahimi , chem . * 46 * , 1977 ( 1991 ) . s. rege and h. fogler , chem . sci . * 42 * , 1553 ( 1987 ) . h. chiang and c. tien , aiche j. * 31 * , 1349 ( 1985 ) . v. g. levich , _ physicochemical hydrodynamics _ ( prentice - hall , 1962 ) . h. e. daniels , proc . london , ser . a * 183 * , 404 ( 1945 ) ; d. g. harlow and s. l. phoenix , j. comput . * 12 * , 195 ( 1978 ) ; s. l. phoenix and r. l. smith , j. appl . mech . * 103 * , 75 ( 1981 ) ; d. sornette and s. redner , j. phys . a * 22 * , l619 ( 1989 ) ; p. l. leath and p. m. duxbury , phys . b * 49 * , 14905 ( 1994 ) . b. kahng , g. g. batrouni , and s. redner , j. phys . a * 20 * l827 ( 1987 ) . g. h. thomas , g. r. countryman , and i. fatt , soc . j. * 3 * , 189 ( 1963 ) ; i. chatzis and f. a. l. dullien , int . eng . , 47 ( 1985 ) . j. koplik , s. redner , and e. j. hinch , phys . e * 50 * , 4650 ( 1994 ) l. niemeyer , l. pietronero , and h. j. wiesmann , phys . * 52 * , 1033 ( 1984 ) ; s. havlin , m. dishon , j. e. kiefer , and g. h. weiss , phys . * 53 * , 407 ( 1984 ) . w. hwang , phd thesis ( boston university , 2000 ) . w. h. press , s. a. teukolsky , w. t. vetterling , and b. p. flannery , _ numerical recipes in c _ ( cambridge , 1999 ) . r. vaidyanathan and c. tien , chem . sci . * 46 * , 967 ( 1991 ) . we solve the master equations , eq . ( [ bubbleme ] ) , by the laplace transform method . define @xmath249 , and take laplace transform of eq . ( [ bubbleme ] ) to find @xmath250 rearranging yields @xmath251 we multiply the above equations for indices 0 , 1 , @xmath252 together and rearrange terms yet again to obtain @xmath253 before taking averages , we use @xmath254(1+\tau_ks)$ ] on the left hand side of eq . ( [ laplace1 ] ) to get @xmath255 averaging over the residence time distribution @xmath122 defined by eq . ( [ ftau ] ) , we obtain @xmath256 where @xmath257 denotes an average over @xmath122 . note that @xmath56 does not depend on @xmath258 because all the @xmath259 s are independent and identically distributed in the large @xmath49 limit . for the long - time limit , we need the small-@xmath260 behavior of @xmath56 . accordingly , we expand @xmath261 in terms of the moments @xmath262 . the profile of the carrier fluid ( dashed line in fig . [ fig1 ] ) is monotonically decreasing near the front , thus can be characterized by its first derivative , which gives a bell shaped distribution centered at the front . we divide the derivative by the total sum of derivatives @xmath263 to obtain the normalized probability distribution of the front @xmath264.\ ] ] the average position of the front is , using eq . ( [ pntilde ] ) , and the steady state solution of eq . ( [ bubbleme ] ) , @xmath265 , @xmath266\nonumber\\ & = & { 1\over\langle p_0\rangle } { \cal l}^{-1 } \bigl(\sum_k k [ \langle\tilde p_{k-1}\rangle-\langle\tilde p_k\rangle]\bigr)\nonumber\\ & = & { 1\over\langle\tau\rangle}{\cal l}^{-1}\bigl({1-b\over s^2 } \sum_k k[b^{k-1}-b^k]\bigr)\nonumber\\ & \simeq&{t\over \langle{\tau}\rangle},\end{aligned}\ ] ] where the over - bar means averaging over @xmath267 , and @xmath268 is the inverse laplace transform . we use the identity @xmath269 for the last step . we assume sufficiently long chain of bubbles in the summations above to prevent finite length effect . from above , we find the propagation velocity as @xmath270 ( eq . ( [ hydrov ] ) ) . similarly , @xmath271 $ ] , and the width of the front is @xmath272^{1\over2 } \simeq\left\{{t\over\langle{\tau}\rangle}\left [ 2\gamma(1+{\mu\over2})\gamma(3-{\mu\over2})-1\right]\right\}^{1\over2},\ ] ] which gives eq . ( [ hydrow ] )	we study the kinetics of _ infiltration _ in which contaminant particles , which are suspended in a flowing carrier fluid , penetrate a porous medium . the progress of the `` invader '' particles is impeded by their trapping on active `` defender '' sites which are on the surfaces of the medium . as the defenders are used up , the invader penetrates further and ultimately breaks through . we study this process in the regime where the particles are much smaller than the pores so that the permeability change due to trapping is negligible . we develop a family of microscopic models of increasing realism to determine the propagation velocity of the invasion front , as well as the shapes of the invader and defender profiles . the predictions of our model agree qualitatively with experimental results on breakthrough times and the time dependence of the invader concentration at the output . our results also provide practical guidelines for improving the design of deep bed filters in which infiltration is the primary separation mechanism . 2
though many are taught that there are merely three phases of matter , _ solid , liquid , _ and _ gas _ , the understanding of broken symmetries , nambu - goldstone modes @xcite , and the anderson - higgs - kibble @xcite mechanism allows prediction , control , and elucidation of other novel forms of matter . liquid crystalline phases interpolate between the simple forms of matter sketched above . the nematic has partially broken rotational invariance @xcite , the smectic phase has broken one - dimensional translation invariance @xcite , and the hexatic phase @xcite has broken two - dimensional rotational invariance . these materials not only afford a deeper understanding of condensed matter , they ( and their lyotropic cousins , diblock copolymers @xcite ) , are also of great technological interest , from displays to coatings , from drug delivery to hybrid materials . technological applications provide challenges for the theorist and , in particular , pose problems with imperfect boundary conditions and geometries . in this paper , we focus on striped or smectic phases on _ curved _ substrates . our work is motivated by elegant experiments by hexemer and kramer @xcite who probed films of diblock copolymers on corrugated surfaces . explaining the properties of columnar layers in this geometry is important for controlling the resulting microstructure of the phase on frozen , undulating surfaces . lamellar , smectic , striped , and columnar phases can all be treated in a single unified , framework on a two - dimensional surface , with the columns , layers , _ etc . _ lying _ in _ the tangent plane . the physics of smectics is beautiful and intricate : the coupling between their geometry and rotation invariance ensures the presence of essential nonlinearities in the strain that lead to anomalous elasticity @xcite and dramatic departures from linear elasticity even for small strains @xcite . these nonlinearities make the theory of the nematic - to - smectic - a transition notoriously difficult as well @xcite and the effort to capture them has enhanced our understanding of global versus local symmetries . posed with complex boundary conditions , smectics often form focal conic domains @xcite in which the layers remain equally spaced but acquire a large curvature to accommodate their growth in a confined geometry . in the past decade or so , smectic phases with cubic order have been discovered @xcite , which have pushed our understanding of the competing tendencies of uniform layer spacing and non - vanishing curvature . indeed , the two requirements can not generically be reconciled with the layer topology @xcite . this leads to a subtle type of geometric frustration in which there is no local obstruction to finding uniformly - spaced layers but their construction leads to diverging layer curvatures . these focal conics are , in fact , analogous to the caustic singularities in optics @xcite . it has been conjectured that the frustration in smectics can be lifted by introducing curvature in the background space @xcite , much as the nematic blue phase is not frustrated on the surface of a four - dimensional sphere @xcite . it is , of course , not possible to experimentally study layered systems in high dimensional space . however , a simpler but not less subtle form of geometric frustration also exists in two - dimensional smectics lying on curved substrates @xcite . in contrast to the case of two - dimensional crystals , @xcite the connection between strain and substrate curvature is indirect , and consequently , configurations with uniformly - spaced layers are numerous . however , the layers nevertheless inherit curvature from the underlying substrate due to compression elasticity . smectics exhibit quite different phenomena from other ordered phases on curved surfaces and new theoretical ideas are required to understand them . in this paper , we study configurations of uniformly - spaced layers on a curved surface , elucidating and elaborating on the mechanisms of geometrical frustration and their effects on layer configurations . in addition , we explicitly consider the effect of the energetic cost of bending the layers to lie on the surface . this extrinsic bending effect leads to a purely geometrical ordering field for the layers . though much effort has recently been directed toward this problem , particularly theoretical @xcite and numerical @xcite , this extrinsic curvature effect has received relatively little attention . in fact , bending energies can play an important role in ordering stripe patterns over longer length scales than would otherwise be possible @xcite . this observation suggests the use of surface curvature as a motif to control self - assembly of block copolymers @xcite . consider , for example , the problem of repulsive semi - flexible polymers on a cylinder . in the fixed density ensemble , the polymers will attempt to adopt an equal spacing . as shown in fig . 1 , there are many such structures ranging from `` pin stripes '' to `` rugby stripes '' , as well as the continuous class of `` barber pole '' textures . what do these have in common ? they all have equally spaced polymers and , from the point of view of _ intrinsic _ geometry , they are all straight , that is , they are all geodesics . this can be seen , for instance , by cutting open the cylinder and laying it flat all of these stripe textures map into straight lines on the flattened cylinder . however , these polymer - like lines are three - dimensional and their energy is a function of their embedded conformation in @xmath0 . the `` pin stripe '' texture in fig . [ fig : cylinders ] is the only one for which the polymers have vanishing three - dimensional curvature and will thus be the ground state . to this energetic accounting , we add a compression energy which favors equal - spacing between the lines and an intrinsic curvature energy within the local tangent plane . we write the total free energy schematically as : @xmath1\nonumber\\&&\quad + [ \hbox{intrinsic bending of lines } ] \nonumber\\&&\qquad+ [ \hbox{extrinsic bending of lines}]\end{aligned}\ ] ] in sec . [ sec : smectic ] we will develop explicit , geometric expressions for these three contributions . however , to frame our analysis , we pause to discuss the subtle frustration which arises from these competing terms . the compression term measures the deviation of the layers from being equally spaced . on a curved substrate it is still possible to define `` equal - spacing '' : at each point on a given line we move along the geodesic tangent to the line s normal for a fixed interlayer distance ( see , for instance , the right side of fig . [ fig : gaussbonnet ] ) . this construction generates equally - spaced layers . however , in contrast to a cylinder , on a curved substrate this is not straightforward . to see this , we consider lines on cones ; the cone is flat everywhere but for its apex where it has a concentration of gaussian curvature . in fig . [ fig : cones ] we show three different smectic complexions on a cone . each is generated by first tracing a smectic pattern on a flat sheet . removing a wedge from the sheet allows us to construct a cone a surface with vanishing gaussian curvature everywhere except at the point . geodesics on the cone become straight lines on the original surface and _ vice versa_. on the left we see equally - spaced lines ( solid ) and their normal flows ( dashed ) . when put onto a cone , the geodesics remain equally spaced . since the intrinsic bending vanishes , by definition , for geodesics , this pattern has no contribution from the first two terms almost everywhere . however , due to the gaussian curvature conentrated at the tip of the cone , the global effect of the gauss - bonnet theorem ( discussed below ) is that a sharp , intrinsic bend must be introduced into the layers along a seam of the cone to account for the angle deficit . moreover , though these lines are flat from the point of view of the surface , they are bent in @xmath0 and thus contribute to the _ extrinsic _ curvature energy as well . since the directions along which the extrinsic curvature vanishes depends on how the surface is bent into three - dimensions , this effect leads to an extrinsic field which can align the layers ( see below ) . it is instructive to consider the other two configurations in fig . [ fig : cones ] as well . in the center figure a set of concentric circles is wrapped onto a cone . from the construction we know that the resulting lines are equally - spaced and have no sharp bends . the circles have _ both _ intrinsic and extrinsic curvature ; we can see this by noting that the normal to each circle points towards its center , in the same plane as the circle . however , the normal to the cone s surface does not lie in that plane and so the circle s curvature has components both along the surface normal and perpendicular to it . the former measures the extrinsic or normal curvature , the latter the intrinsic or geodesic curvature . finally , on the right , we show a set of lines which all converge to a point . they are not equally - spaced on either the plane or the cone ; by construction they are geodesics and for this particular embedding of the cone , they have vanishing extrinsic curvature . the simple cone geometry highlights the geometric frustration embodied in simultaneously minimizing the three competing terms in the free energy . in section ii we introduce the geometric tools necessary to characterize stripes and develop the free energy that controls smectic order on curved surfaces . in section iii we present a number of examples that illustrate the effects of both intrinsic and extrinsic geometry on striped pattern formation . because of the many different limits that might be considered in this frustrated system , we focus primarily on zero - strain complexions where the lines are equally spaced ( compression energy dominates ) . this limit is closely related to the geometric optics limit of light propagation , as we will exploit . as we shall see , even with this constraint the competition between the remaining two terms is quite subtle . in section iv we discuss local mechanisms ( related to extrinsic curvature energy ) which lead to long range , two - dimensional order on curved surfaces , bypassing the standard coleman - mermin - wagner conclusions about thermal fluctuations in flat space @xcite . finally , we summarize our results and discuss open questions . given the focus in general relativity and string theory on intrinsic geometric concepts such as conformal invariance , diffeomorphism invariance , and modular invariance , it is worth remembering that fascinating materials exist in three dimensions where not only intrinsic but also _ extrinsic _ geometry plays a key role . with extrinsic curvature in mind , recall that an embedded surface @xmath2 has an induced metric @xmath3 ( first fundamental form ) , a unit surface normal @xmath4 , and a curvature tensor @xmath5 ( second fundamental form ) : @xmath6 where the last equality follows from differentiating @xmath7 . note that our sign convention for @xmath5 reflects the standard sign for the curvature of a three - dimensional space curve , _ i.e. _ a circle will have a positive curvature when its normal is chosen to point _ inward_. here and throughout we refer the reader to @xcite for more technical details . the two principle curvatures @xmath8 and @xmath9 are the eigenvalues of @xmath10 , where the indices associated with @xmath11 and @xmath12 are suppressed to simplify the notation . the unit eigenvectors associated with the matrix @xmath13 are the principal directions @xmath14 and @xmath15 . on a general surface the curvatures and directions change from point to point . with stripes , polymer strands , or smectic layers in mind , consider a curve embedded in the surface , @xmath16 , with @xmath17 measuring the arclength . the curve has unit tangent @xmath18 with derivative @xmath19 an equation which defines the unit normal to the curve , @xmath20 , and the curvature , @xmath21 . because @xmath22 and the surface normal @xmath23 defined above are not necessarily at right angles , we write @xmath24 so that the change in @xmath25 is decomposed into a vector along the surface normal , @xmath4 , and a vector in the surface , @xmath26 . the coefficients @xmath27 and @xmath28 are the normal and geodesic curvature , respectively , and obey @xmath29 . a geodesic on a surface is a curve for which the change in the tangent vector has no components in the surface , that is @xmath30 . however , even a great circle on a sphere curves in three - dimensional space . at any point on a curved surface , we can choose coordinates so that @xmath31 are orthonormal so that @xmath32 and @xmath33 where @xmath34 indicates the transpose @xcite . it follows that the normal curvature can be further decomposed as @xmath35 where @xmath36 gives the angle @xmath37 between the curve and one of the principle directions . the last equality in the first line follows from the definition of @xmath13 in ( [ eq : geometrydefs ] ) and from expanding @xmath38 in terms of the two tangent vectors @xmath39 @xcite . on a radially symmetric surface ( such as the gaussian bump discussed below ) with height function @xmath40 , @xmath41 $ ] . the principal directions are , by symmetry , along @xmath42 and and along @xmath43 . it follows that if @xmath37 is the angle between the curve and the radial direction , then @xmath44 where @xmath45^{3/2}}\nonumber\\ \kappa_\theta & = & \frac{\partial_r h}{r \left[1+\left(\partial_r h\right)^2\right]^{1/2}}\label{eq : kappa}\end{aligned}\ ] ] when @xmath46 ( on the flanks of a gaussian bump ) , there will always be an angle @xmath37 for which @xmath27 vanishes , and so the extrinsic geometry and extrinsic bending energy of stripes on a surface set a natural , local direction for the polymers . now our discussion for smectic layers on cylinders ( recall fig . [ fig : cylinders ] ) can be made precise . on a cylinder of radius @xmath47 , @xmath48 and @xmath49 so @xmath50 only when @xmath51 , which selects the preferred minimum energy configuration . the cylindrical ground state is unambiguous because all of the textures in fig . 1 are composed of lines drawn along geodesics . note that this _ extrinsic _ effect is sensitive to an _ intrinsic _ quantity , the angle @xmath37 . on a smooth bump , with a nonzero , spatially varying gaussian curvature , the problem is complicated by the intrinsic geometry of the curves defined by the layers . now we must also consider @xmath28 , the geodesic curvature . as we will see in the following , the surface geometry may not only favor line configurations for which @xmath52 but , more importantly , the surface geometry can prevent the stripes from achieving both equal spacing and vanishing @xmath28 . lamellae , both on surfaces and in the plane , are conveniently represented by the level sets of a function @xmath53 , where @xmath54 is the equilibrium layer spacing and @xmath55 is an integer labeling the layer . the phase field @xmath56 is related to the average mass density @xmath57 of the layers by @xmath58 . the gradient @xmath59 contains information both about the layer spacing , related to its magnitude , and the direction normal to the layers . we will use @xmath60 to denote the covariant derivative on the surface , defined as @xmath61 for a scalar function @xmath56 , and @xmath62 for ( contravariant ) vectors @xcite . when unambiguous , we will also use @xmath63 to indicate contraction : thus , @xmath64 . note that @xmath65 gives the change in @xmath66 when one moves in an arbitrary direction @xmath67 on the surface . upon specializing to the case where @xmath68 , the resulting change is @xmath69 . hence , equal layer separation implies the condition @xmath70 . the strain , @xmath71 , measures the deviation from equal spacing . while there are many possible forms for @xmath71 , depending on microscopic details , we only require that @xmath71 vanish when @xmath72 and , in the small deformation limit where the eulerian displacement is @xmath73 , that @xmath74 where @xmath75 is the local direction of the layer normal . a suitable form is @xmath76 the energy cost of small deformations from equal spacing will be an expansion in powers of @xmath71 , with the lowest order term being @xmath77 . if the layers are equally spaced , then @xmath78 and , upon differentiation , we find that @xmath79 where @xmath80 is normal ( within the tangent plane ) to the curve used to define a layer . the condition @xmath81 is precisely the geodesic equation for the layer normal . we see then , independent of the form of the strain , when @xmath82 ( thus , minimizing the strain energy ) the layers are spaced evenly along geodesic curves . if we consider only zero strain solutions , this condition reduces the problem of finding smectic textures to a _ first - order _ evolution equation . thus , the normals lie on `` straight lines '' on the surface ( _ i.e. _ geodesics ) ; layer curvature leads to their convergence into singularities . indeed , if @xmath83 measures the distance between two geodesics along a third , perpendicular geodesic , the best local approximation to parallel trajectories leads to @xcite @xmath84 where @xmath85 is the gaussian curvature . this geodesic deviation equation naturally reminds us of geometric optics @xcite , in which the layer normals act as rays and the gaussian curvature , @xmath86 as a variable index of refraction . in regions of negative curvature the normals diverge as through a diverging lens ; in regions of positive curvature they converge . converging patterns of rays form caustics ; in the language of ordered lamella , caustics are curvature singularities in the lines . on a curved surface , a set of geodesics normals initially perpendicular to some layer may cross a finite distance away , leading to a cusp - like boundary where the geodesics finally converge . this system thus provides a low - dimensional analog of gravitational lensing @xcite . cusps are regions of high bending energy and thus , in the true ground state , they will be smoothed out at the cost of compression energy . we are now in a position to formulate the complete free energy : @xmath87,\ ] ] where @xmath88 is the bulk modulus and the couplings @xmath89 and @xmath90 are length scales , typically on the order of a column diameter , which measure the relative importance of bending and compression energies . in principle @xmath91 : the @xmath92 limit is the `` generally covariant '' limit where extrinsic effects are irrelevant . on the other hand , if we were trying to decorate the surface with semi - flexible polymers with persistence length @xmath93 , we would have @xmath94 , where @xmath95 is the thickness of the polymer layer normal to the surface and @xmath54 is the average spacing between the polymers @xcite . in general , nonzero @xmath96 and @xmath97 will lead to corrections to the simple picture of caustic singularities sketched above . to better understand eq . ( [ eq : energy ] ) , consider the three different decorations of the cone shown in fig . 2 , assuming finite cones of radius @xmath47 . on the left , while there is some extrinsic layer bending leading to an energy on the order of @xmath98 , the dominant energetic cost arises from the bend wall on the seam : there the equal - spacing condition will breakdown in order to smooth out the sharp kink . equivalently , we can view this as a row of dislocations as arises in low - angle grain boundaries with elastic core energy per unit length scaling as @xmath99 . in either case , we see that the energy of the grain seam scales as @xmath100 . upon focussing on the central cone , we see that the smectic complexion has energetic contributions from both bending moduli and @xmath101 . consider finally the cone on the right ; the texture there has no bending energy at all , but a very large has compression energy , @xmath102 . from this rough analysis , we would expect that away from regions of large gaussian curvature ( in this case the conical singularity ) , the `` bulls - eye '' pattern of the center cone of fig . 2 will be the dominant line texture . however , depending on the various elastic constants this can break down for finite regions around the the peak . in the following we will consider these effects , in particular in section iv . for further insight into eq . ( [ eq : energy ] ) , it is instructive to consider this free energy in the context of weakly deformed two - dimensional smectics coating a cylindrical surface with a flat metric and no singularitiy . we introduce the usual one - dimensional displacement field @xmath103 and consider the simple case of a cylinder discussed in the introduction . in the coordinate system of fig . [ fig : gaussbonnet ] , with @xmath104 , we take as our level set function @xmath105 consider the limit of gentle , slowly - varying undulations superimposed on the texture shown at the bottom left of fig . [ fig : gaussbonnet ] . for a cylinder of radius @xmath47 , it is straightforward to show that the free energy then becomes @xmath106,\ ] ] where @xmath107 is the local tilt angle that the stripes make with respect to the @xmath108-axis . the last term in eq . ( [ eq : nelson2 ] ) follows from eq . ( [ eq : kappan ] ) , where we choose axes of principle curvature such that @xmath109 and @xmath110 . in addition to the usual terms describing a rotationally - invariant 2d smectic @xcite , there is now a weak ordering field proportional to @xmath111 . the field is weak in two senses its strength is proportional to @xmath112 and hence vanishes in the limit of a cylinder of infinite radius . furthermore , it goes like @xmath113 for small tilt angle deviations from the preferred direction . nevertheless , this field is enough to break the symmetry and give rise to a preferred direction for the lines . as discussed above , this field arises from the extrinsic curvature tensor and will vary spatially in both magnitude and direction on more general curved surfaces . in the remainder of this paper , we first neglect this weak field ( for a gaussian bump of height @xmath114 and size @xmath115 , @xmath116 replaces @xmath117 ) and focus on satisfying the constraint of equal layer spacing imposed by the first term of eq . ( [ eq : energy ] ) , with a simplifying boundary condition at infinity . we then discuss effects due to this extrinsic ordering term with free boundary conditions at infinity . an anisotropic membrane would have two bending moduli , one for bends along a `` hard '' direction , the other for bends along the `` soft '' direction . if the anisotropy had a vector character , we could also include a cross term , allowed by symmetry . defining @xmath118 as the unit vector lying along the layers , an alternate form for the normal curvature is @xmath119 , where @xmath12 is the surface curvature tensor given in equation ( [ eq : geometrydefs ] ) . the other two allowed terms are @xmath120 and @xmath121 , respectively . the magnitude of the moduli for these additional terms depend on the molecular details of the smectic layers . these elastic constants are different , in principle , from the intrinsic bending modulus @xmath122 in ( [ eq : energy ] ) . for concreteness , we will focus our discussion on a monolayer of block copolymer cylinders lying on the surface . the columnar phase of neat ( _ i.e. _ monodisperse with no solvent ) block copolymers , though bearing resemblance to the columnar hexagonal phase of liquid crystalline polymers , is actually an incompressible three - dimensional elastic medium . strong - segregation calculations @xcite suggest that the columns can be viewed as semi - flexible rods . in that case `` bending along the columns '' is the `` hard '' direction , being more energetically costly than `` bending perpendicular to the columns '' for a few - layer coating of a curved substrate . roughly speaking , the columnar phase is similar to a corrugated sheet with the columns corresponding to the corrugations . however , as the diblock film grows , bending along the columns leads to deformations that are independent of the thickness since lamella can slide past each other with no cost . on the other hand , bending perpendicular to the columns requires large amounts of crystalline strain . for a thin film composed of only a few layers , the bending energy we consider here should dominate . however , how the introduction of low - angle grain boundaries and surface energies affects these calculations is an open question @xcite . to keep our analysis from becoming highly ramified , the only non - vanishing extrinsic elastic modulus will be bending along the column tangents . with the spate of recent work on both crystalline order and nematic order on curved substrates , it is valuable to compare and contrast these systems with the smectic a phase with one - dimensional crystalline order that lives _ between _ the crystal and the nematic . we shall see that the smectic presents issues all its own and affords a fresh arena for the interplay between geometry and soft materials . a nematic liquid crystal on a surface is described by a unit vector @xmath123 which lives in the tangent plane of the surface . the standard three - dimensional frank free energy @xcite @xmath124 & = & \frac{1}{2}\int d^3\!x\,\left\{k_1\left(\nabla\cdot{\bf n}\right)^2 + k_2\left[{\bf n}\cdot\left(\nabla\times{\bf n}\right)\right]^2\right.\nonumber\\ & & \qquad\left . + k_3\left[\left({\bf n}\cdot\nabla\right){\bf n}\right]^2\right\}\end{aligned}\ ] ] is modified in two ways : first , @xmath125 when @xmath126 lies on a smooth , two - dimensional surface and depends only on the two surface coordinates @xmath127 . the two - dimensional nematic free energy then reads @xmath128 ^ 2,\ ] ] where @xmath129 and @xmath130 is the determinant of the metric tensor @xmath3 . the first term in eq . ( [ eq : nematic ] ) penalizes director splay , whereas the second term penalizes director bend . when @xmath131 the free energy in eq . ( [ eq : nematic ] ) is isotropic and can be cast in a form that naturally lends geometric insight . consider a local angle field @xmath132 , corresponding to the angle between @xmath133 and an arbitrary , orthonormal local reference frame whose basis vectors we label @xmath134 with @xmath135 . in the one frank constant approximation , the free energy in eq . ( [ eq : nematic ] ) can be recast in the form @xmath136 where @xmath137 is a connection that plays the role of the christoffel symbols by compensating for the rotation of the 2d basis vectors @xmath134 in direction @xmath138 and @xmath139 @xcite . since the curl of @xmath137 is equal to the gaussian curvature @xmath140 , the nematic energy can not vanish on a surface with non - zero gaussian curvature . note that @xmath137 is , in general , a non - conservative field , so we can not minimize ( [ eq : patic - ener ] ) by setting @xmath141 equal to @xmath142 everywhere on the surface . this property is a manifestation of a more general mechanism , commonly referred to as geometric frustration , to indicate situations where the molecular arrangement favored by local interactions can not be extended globally . as the gaussian curvature of the substrate increases , defects are generated in the ground state to lower the energy cost of geometric frustration . their energetics is analogous to coulomb particles interacting with a smeared out electrostatic charge given by the gaussian curvature . this nontrivial result can be rationalized by examining the free energy in eq . ( [ eq : patic - ener ] ) and noticing that the connection @xmath137 and the gaussian curvature @xmath140 are analogous ( in _ two _ dimensional electromagnetism ) to a frozen vector potential and the magnetic field respectively . the topological defects , _ i.e. _ disclinations , behave as monopoles in the dynamical field @xmath132 whose interaction with the geometry of the surface is mediated by the geometric gauge field @xmath137 . an additional coupling between defects and the metric arises from the metric factors that appear , for example , in the surface element @xmath143 independently of the connection @xcite . the physics of geometric frustration is at work also in the more complicated setting of curved space crystallography @xcite . now the orientational order of the nematic is supplemented by translational degrees of freedom . for gently deformed surfaces , the crystalline energy can be expressed in terms of the lam coefficients @xmath144 and @xmath145 @xcite @xmath146 where @xmath147 are cartesian coordinates in the plane and @xmath148 $ ] is the strain tensor . compared to its flat space counterpart , the strain tensor has an additional term @xmath149 that couples the gradient of the displacement field @xmath150 to the gradient of the surface height function @xmath151 . the field @xmath152 is a tensor version of the connection @xmath153 introduced above to describe orientational order on curved surfaces . indeed the curl of the tensor field @xmath152 is equal to the gaussian curvature of the surface @xmath154 where @xmath155 is the antisymmetric unit tensor ( @xmath156 ) @xcite . by the same reasoning as before , the integrand of eq . ( [ eq : lame ] ) and hence the ground state energy can not be made to vanish . this is the mathematical mechanism by which geometric frustration enters the physics of curved crystals . it can be grasped more intuitively by recalling that bending a plate into a surface of non vanishing gaussian curvature necessarily causes it to stretch @xcite . smectic liquid crystals , on the other hand , can maintain uniform layer spacing ( and hence achieve zero strain ) even in the presence of gaussian curvature . the gaussian curvature , nevertheless , couples to the curvature of the layers . to see how this coupling appears , we start by noticing that the bend coupling proportional to @xmath157 in equation ( [ eq : nematic ] ) resembles the geodesic equation for curves tangent to the director . we can establish an intuition for smectic patterns , therefore , by studying nematics with very large @xmath157 , and identifying the nematic order parameter as the layer normal . nematics on spheres are required to have a net @xmath158 topological charge , which tend to break up into four @xmath159 disclinations . when @xmath160 , the disclinations lie on the corners of a tetrahedron @xcite ; when @xmath157 is large , however , the @xmath159 disclinations lie on an equator @xcite and the local texture is a lines of longitude structure . at the other extreme , when @xmath161 is large the director takes on a lines of latitude texture . since there is no difficulty to finding geodesics in any direction at any point on a curved surface , there is no local obstruction to constructing equally - spaced layers on any surface . however , doing so while simultaneously finding layers which are also geodesics is impossible . this can be seen by constructing a rectangle with two opposite sides given by adjacent layers and the remaining two sides given by geodesics normal to both layers ( see fig . [ fig : gaussbonnet ] ) . upon applying the gauss - bonnet theorem @xcite to this contour , we obtain @xmath162 where @xmath163 is the turning angle at each corner of the rectangle , we find @xmath164 the integral on the left - hand side is over the area of the square while the integrals on the right - hand side are over the two adjacent layers . since @xmath165 in general , it is not possible for @xmath166 on both layers . despite our local ability to set @xmath82 in equations ( [ eq : straindef ] ) and ( [ eq : energy ] ) , there may be global obstructions to finding low energy configurations with vanishing compression strain . to see this , we first recast ( [ gb ] ) locally . upon using the local coordinates @xmath167 and introducing the surface metric @xmath3 , eq . ( [ eq : gb2 ] ) becomes @xmath168 where @xmath169 is the derivative along the normal direction and we have used the fundamental theorem of calculus . after rewriting the integral on the right in the local coordinates , we come to the local relation @xmath170 recall that the geodesic curvature is the fractional rate of change of the length of an arc as it is moved perpendicular to itself and that positive curvature implies that normal evolution shrinks the curve @xcite . it follows that @xmath171 where we define the normal direction to be along the curve s ( inward ) pointing normal . using eq . ( [ eq : gb3 ] ) , we find @xmath172 . dividing through by @xmath173 , we find @xmath174 this evolution equation encapsulates the geometric frustration implicit in the gauss - bonnet theorem . indeed , following the same reasoning , eq . ( [ eq : divergence ] ) is also a local form of gauss - bonnet ; we may take an arc of length @xmath83 connecting the two diverging ( converging ) geodesics at @xmath17 . then the curvature of those arcs is @xmath175 , and it follows that @xmath176 , which is identical to ( [ eq : evolution ] ) . in this section , we review and generalize our prior results on smectic textures on a simple gaussian bump @xcite to more complex geometries . consider smectic order on a curved substrate described in the monge representation by a height function @xmath177 ; the gaussian curvature is then given by @xcite @xmath178}.\ ] ] if the whole surface can be described by @xmath177 then it is topologically equivalent to the plane and , by the gauss - bonnet theorem , @xmath179 . we start with the gaussian bump used to study nematic order in ref . @xcite , @xmath180 , with @xmath181 . @xmath85 is positive near @xmath182 and negative for large @xmath183 . at an intermediate radius @xmath184 , @xmath85 vanishes . a particularly simple smectic configuration results from choosing radial geodesics as the layer normals , in which case we would find uniformly - spaced , azimuthal layers , a disclination at the top of the bump , and a power law decay of @xmath28 away from the center . is it possible , however , to generate a configuration _ free _ of topological defects on a bump ? suppose we start with straight layers at @xmath185 , so that the layer function @xmath186 defined in sec . [ sec : smectic ] obeys @xmath187 as @xmath188 . this boundary condition would describe an experimental setup where the layers of a diblock columnar phase grow along a temperature gradient parallel to @xmath189 . we expect layer by layer growth nucleated from a boundary at large negative @xmath108 . the geometry of the substrate leads to the formation of singularities , but rather different than the isolated disclination with concentric circular layers centered on the bump discussed above ( see fig . [ fig : layers ] ) . we call these singularities caustics because , just as in geometrical optics , these are places where many initially parallel light rays or , in our case , geodesics converge . at these locations the value of @xmath56 is well - defined but @xmath190 is discontinuous . though it might be tempting to call these cusps `` _ defects _ '' , they are not . recall that near a dislocation the smectic order parameter vanishes and the phase field @xmath56 takes on _ all _ values around the defect . similarly , in the vicinity of a disclination , the nematic order vanishes and the layer normal takes on all directions at the defect . in contrast , the cusp singularities in fig . [ fig : layers ] have definite values of @xmath56 and @xmath190 does not wind through all possible directions around these singularities . thus , although there is a discontinuity in @xmath190 , it is not of the same nature as that of a disclination . in the analogy to optics , we would say that a dislocation is a place of vanishing amplitude , while a caustic is a location of very high , if not infinite , amplitude @xcite . figure [ fig : layers ] shows a birds - eye view of a gaussian bump coated with stripes , with the geodesics as dashed lines and the layers themselves as solid lines . the red circle indicates the locus of points for which @xmath191 and we have shaded the regions according to the magnitude of @xmath28 . we could also have predicted these cusps by determining where the curvature of the smectic layers diverges along the geodesics defined by the layer normal . to do this , we integrate ( [ eq : evolution ] ) along the normal geodesic passing over the top of the bump . we find that @xmath28 diverges a finite distance past the center of the bump , shown in fig . [ fig : layers ] ; this divergence indicates the onset of the infinite curvature cusps in the layer lines . from a more global perspective , the cusp angle is a measure of the integrated gaussian curvature . to see this , we again use the gauss - bonnet theorem to describe evolution of the cusp angle as a function of distance from the center of the bump . the key is to define a geodesic triangle with one edge along the @xmath108-axis , another edge along the layer parallel to the @xmath192-axis at @xmath188 and a third edge along a normal curve , which is a geodesic by construction . the resulting geodesic triangle has two exterior ( or interior ) @xmath193 angles . denote the remaining internal angle by @xmath194 . the corresponding external angle , @xmath195 , is constrained by the gauss - bonnet theorem , which leads to @xmath196 = \alpha\ ] ] where the integral is taken over the inside of the triangle . notice that , as the cusp location @xmath108 becomes large and positive ( see fig . [ fig : layers ] ) , @xmath197 since the triangle incorporates an increasing amount of area in the entire upper half - plane . because the integrated gaussian curvature is zero in the upper half - plane , the angle @xmath194 necessarily decreases with increasing @xmath108 . though the grain boundary persists infinitely far to the right of the center of the bump , the cusp angle asymptotically vanishes . in the case of a gaussian bump , the angle will fall off as @xmath198 , though the details will vary for other surfaces . similar to geometric optics , the formation of these cusps should not depend sensitively on the exact geometry of the substrate ; rather it is a function of the topology as characterized by the intrinsic curvature . we have considered other substrates , all topologically equivalent to the plane so that again , the integrated gaussian curvature vanishes , @xmath199 . in fig . [ fig : x2plusy2 ] we depict both the geodesics and layers generated from the same boundary condition as @xmath200 that we considered in the last section ; now , however , we have chosen a more complex axisymmetric bump with multiple regions of positive and negative @xmath85 . note that geodesics that do not go through the central region of the bump , outlined by the outermost red circle , will generate a set of cusps at large @xmath108 along the midline similar to those in fig . [ fig : layers ] . however , we now find that there are additional caustic lines that form as a result of the focussing from the inverted dimple of the bump . these `` fold '' caustic lines , in this case , converge toward the @xmath108-axis and eventually end there . to see this , we adapt the discussion in the last section : geodesics that avoid the central region entirely lead to larger - angle cusps on the @xmath108-axis than those that go through the same region . at large @xmath108 , the pattern must be that of the simple bump in fig . [ fig : layers ] and thus there will be no auxiliary caustics at large @xmath108 and @xmath192 . it follows that any extra caustics must converge to the @xmath108-axis . we can see this in another way by considering the focussing equation ( [ eq : divergence ] ) . pairs of geodesic normal lines that remain in the outer region always diverge , because @xmath201 @xcite . however , a pair of geodesics , one of which remains in the outer region and which enters the inner region will still diverge less slowly and , indeed , pairs which both travel through the annulus with @xmath202 will _ converge_. hemmed in by the pairs that are always diverging , we see that these additional cusps must form and that the `` outer '' geodesics overtake the `` inner '' geodesics , leading to two extra cusp line grain boundaries , one of which is shown in the lower half of fig . [ fig : x2plusy2 ] . just as microscopists design and choose multiple lenses to manipulate an image , one might hope to engineer a surface whose geometry leads specific smectic textures . to further our intuition , we have studied the non - axisymmetric saddle - bump shown in fig . [ fig : x2minusy2 ] , with the polynomial @xmath203 multiplying a gaussian envelope . as shown in the inset , this surface has a saddle in the center , surrounded by four lobes of positive curvature . upon choosing the same boundary conditions as @xmath188 as before , we find a line of cusps along the @xmath108-axis and two extra off - axis caustic lines that disappear as they approach @xmath204 . in order to find an example for which the auxiliary lines _ diverge _ ( a diverging instead of converging lens ) , we consider , for instance , the same saddle - bump structure , but rotated @xmath205 so that the geodesics that start near the @xmath108-axis minimize their transit across regions of positive curvature . these geodesics are always diverging , picking up only a small amount of @xmath202 as they exit the central saddle region . as a result , there are no cusps on the @xmath108-axis the layers are perfectly smooth since the geodesics associated with the layer normals are not focused to this line . as shown in fig . [ fig : xy ] , one can see that the cusps in the layers form symmetrically , off - axis and that , for the bump shown , they bend away from the line @xmath204 ; it would be impossible for them to converge since the layers are regular on the @xmath108-axis . it seems likely that , by adjusting the saddle - bump parameters , the lines of cusps could be made parallel to @xmath206 . these simple , high - symmetry cases provide intuition for how layer focussing could be used to construct desired patterns far from the bump . it is amusing to consider other boundary conditions ( for instance , a circular or parabolic arc at infinity incident on a bump with little or no symmetry . certainly the `` optical '' elements that we have studied here could , themselves be used to control the layers and geodesic normals if they are put far enough apart ; the bump in fig . [ fig : x2plusy2 ] acts as a converging lens , while the bump in fig . [ fig : xy ] acts as a diverging lens . the effect of such a compound `` lens '' can be seen in fig . [ fig : compound ] . the converging bump on the left leads to the formation of an @xmath108-axis , downstream grain boundary , as it does in fig . [ fig : x2plusy2 ] ; this particular the grain boundary ends due to the diverging effect in the negatively - curved center region of the bump from fig . [ fig : xy ] , but two new diverging grain boundaries appear downstream of this element . we look forward to a future where `` smectic optometrists '' manipulate the layered order with gaussian curvature to make novel devices and materials . in currently available experimental data @xcite , the stripes form all at once across the surface as the block copolymers cure . similarly , if a nematic on a substrate were cooled into the smectic phase , the formation of striped order might not proceed from left to right as imagined above unless a temperature gradient or a strongly nucleating boundary condition at @xmath188 were imposed . moreover , in the prior subsections , we have only considered the effect of the intrinsic geometry , embodied in the gaussian curvature , on the formation and orientation of the layered structure . this approximation amounts to taking the limit @xmath97 and @xmath96 to zero in eq . ( [ eq : energy ] ) . we know from our discussion in the introduction that , when the bending couplings @xmath97 and @xmath96 in eq . ( [ eq : energy ] ) are important , the layers prefer to be perpendicular to the lines of @xmath191 in figs . [ fig : layers]-[fig : xy ] . can a smectic texture with equally - spaced layers be found so that it agrees with the preferred direction along lines given by the @xmath191 locus ? in this section we will consider the additional effect of the normal curvature coupling @xmath97 but set @xmath207 . the intrinsic curvature term differs essentially from the other two in ( [ eq : schem ] ) ; equal spacing is a local constraint and the extrinsic curvature generates a local preferred direction as illustrated in eq . ( [ eq : nelson2 ] ) . however , the intrinsic or geodesic curvature can always be set to zero _ locally _ because there is a geodesic pointing in every direction at each point of the surface . as illustrated in fig . [ fig : gaussbonnet ] it is only after evolution normal to the stripes that the geodesic curvature builds up from zero . thus , the geodesic curvature term does not set a _ local _ directional contstraint . for this reason we neglect @xmath96 in the following discussion and concern ourselves with the orienting effect arising from @xmath97 . presumably , local ordering fields would dominate the kinetics , particularly in layer - by - layer growth . for simplicity we return to the simple gaussian bump in fig . [ fig : layers ] . first , consider the region for which @xmath201 . recall from the discussion around eq . ( [ eq : kappa ] ) that there always exists a critical angle @xmath37 relative to the principal directions such that @xmath27 in eq . ( [ eq : energy ] ) vanishes . as discussed in secs . i and ii , the layers will point in the radial direction along the ring where @xmath191 to minimize the normal bending energy . remarkably , it is possible to construct an equally - spaced array of lines satisfying this boundary condition with free boundary conditions at infinity . this texture is given by the involutes of the ring of zero gaussian curvature , generalized to a curved surface . recall that in flat space , the involutes of a closed plane figure can be generated by wrapping an inextensible string many times around the figure , attaching a pen to the end , and then unwrapping the string under tension @xcite . in optics , the evolutes are the wave fronts which form a caustic the patterns which we have shown in figs . [ fig : layers]-[fig : xy ] are the evolutes , generalized to a curved surface , of the straight line at @xmath185 . the generalization is straightforward : we now require that the string lie within the surface so that distances are measured intrinsically . we can make this observation more precise by finding a family of geodesics normal to the layers , from which we can generate uniformly - spaced stripes . on surfaces of revolution , these geodesics are characterized by clairaut s theorem @xcite , @xmath208 for a constant @xmath209 , where @xmath210 is the angle the geodesic makes with respect to the radial direction . the layers , which lie perpendicular to the normals , therefore make an angle @xmath211 with the radial direction so @xmath212 . as for the bumps discussed above , @xmath183 is the distance from the axis of revolution ; that is , the radial projected distance . when @xmath213 , corresponding to the circle @xmath191 of the gaussian bump , @xmath214 and the layers are radial ; when @xmath215 , @xmath216 and the layers are azimuthal . thus , in the absence of defects , the extrinsic alignment forced at @xmath191 leads to a circular `` lines of latitude structure '' for the layers as @xmath217 [ fig : spiral ] we show these involutes drawn on the gaussian bump of fig . [ fig : layers ] . we also note that just as the layer curvature diverges on the caustic cusps in figs . [ fig : layers]-[fig : compound ] , the curvature of the layers diverges on the ring . to see this directly , calculate @xmath218,\ ] ] where the layer normal has components @xmath219 and @xmath220 . for the layers defined above , this yields , @xmath221 which diverges as @xmath222 . note that our involute _ ansatz _ interpolates between the minimizer of normal curvature near @xmath191 and the low energy , azimuthal `` bulls - eye '' pattern of the central cone in fig . [ fig : cones ] . though it might be natural to assume that the bulls - eye pattern would minimize the overall energy , it is frustrated by the orienting normal curvature energy at the ring of @xmath191 experiment also shows that the layers lie perpendicular to this ring @xcite . certainly in the case of the involutes , the diverging curvature at @xmath191 will be softened by breaking the zero compressional strain condition . from dimensional analysis one can argue that at length scales longer than @xmath97 and @xmath96 , the strain energy dominates the curvature energy and that the @xmath82 approximation is valid . however , at these cusp - like singularities , the curvature always dominates the energy no matter the scale of @xmath97 and @xmath96 . the singularities at these cusps are relaxed by the competition between compression and curvature elasticity . the proper analysis of the breakdown of equal spacing requires a more complete theory which also allows the formation of edge dislocations in the smectic order @xcite . equation ( [ eq : kappan ] ) , when inserted into eq . ( [ eq : energy ] ) , endows the layers with a preferred directionality by minimizing @xmath223 , and , hence , the normal curvature energy . in regions of positive curvature , the minimal energy direction lies along the principal direction with the smallest curvature . in regions of negative gaussian curvature , @xmath224 , so we can solve equation ( [ eq : kappan ] ) for @xmath225 , finding a preferred angle @xmath226 given by ( see eq . ( [ eq : kappan ] ) ) @xmath227 it is instructive to again consider the case of a cylinder of radius @xmath47 , in which @xmath228 along the cylinder axis and @xmath110 . the preferred angle is @xmath229 which implies that the layers all lie along the cylinder axis . why this should be so is immediately apparent from figure [ fig : cylinders ] : when the layers lie along the cylinder axis they are straight in three dimensions whereas when they are azimuthal , each layer has curvature @xmath117 . on our gaussian bump , @xmath180 , and there is a ring of @xmath230 ; near this ring , the surface is cylinder - like and we should expect a preferred radial direction for the layers as discussed above . outside the ring , @xmath231 , and the preferred direction follows equation ( [ eq : normaldirection ] ) in a surface dependent way . there is no general behavior that can be inferred via the expressions for the curvatures in eq . ( [ eq : kappa ] ) when @xmath201 the precise form of @xmath40 dictates the ordering direction . even far from the bump there is no universal direction ; any angle is possible depending on @xmath232 . to probe whether these low curvature directions are actually adopted by the layers , we compute the normal bending energy cost for deviation from the preferred direction for a small patch of layers from fig . [ fig : spiral ] in the region where @xmath201 . a straightforward calculation yields , @xmath233 when @xmath234 . we have used the condition @xmath50 and inserted the preferred angle @xmath226 in order to simplify eq . ( [ eq : deltae ] ) ; the second equality follows from trigonometric identities . the magnitude @xmath235 decreases as @xmath215 but is order @xmath236 up to @xmath237 . at small angles , however , @xmath238 is small and the ordering field is particularly weak . near regions where @xmath239 it follows that @xmath48 and @xmath240 thus there is still a weak , anharmonic ordering field even as the layers approach the ring of zero gaussian curvature , just as for the cylinder described by eq . ( [ eq : nelson2 ] ) . we have outlined a theory of uniformly - spaced lines on a curved surface and used this to understand smectic liquid crystal textures on curved substrates . we started with the more general free energy of eq . ( [ eq : energy ] ) , but specialized to the case where the compression energy dominated ( @xmath241 ) . we then studied the effect of the weak ordering field embodied in a small @xmath242 . this theory is geometrical in nature , and demonstrates that substrate gaussian curvature subtly frustrates smectic order by bending layers and leading to defects and grain boundaries . we have neglected kinetic effects @xcite focussing on ground state configurations . we now comment on additional limitations of our approach for real experimental systems . in particular , two - dimensional smectic order is destroyed by thermal fluctuations and the nucleation of defects @xcite . the remaining intermediate - range order leaves a labyrinthine phase of stripes with anisotropic correlation lengths @xmath243 and @xmath244 that can be considerably larger than a layer spacing , as occurs on a flat substrate . thus , intrinsic curvature effects ( the only effect possible in a flat system ) can be weak , and are likely to be less important than extrinsic curvature in determining layer structure . when the effects of normal curvature are compatible with uniform spacing , we expect the layers to be very well - ordered over long length scales precisely because the curvature acts as an `` ordering '' field . this induced intermediate - range order on a corrugated surface does indeed seem evident in experiments of block copolymer cylinders on a bump @xcite . in particular , the layers will transition from being radial near the radius of @xmath191 to azimuthal far from the bump . where bias introduced by normal curvature disagrees with the constraints required by uniform layer spacing , we expect there will be defects in the layers that accommodate the extrinsic curvature effects more closely . therefore , a more quantitative understanding of the role of defects is necessary to build a complete picture of the ground state of smectic liquid crystals on curved surfaces @xcite . our results can serve as a minimal template for smectic textures from which a more detailed theory of defects can emerge . + we acknowledge stimulating discussions with a. hexemer , g. grason , b. jain , and e. kramer . rdk acknowledges the aspen center for physics , where some of this work was completed . rdk and vv were supported by the national science foundation through grant dmr05 - 47230 , via the penn mrsec grant dmr-0520020 and a gift from l. j. bernstein . work by drn was supported by the national science foundation , through grant dmr-0654191 and via the harvard materials research science and engineering center through grant dmr-0820484 . cds was supported by the national science foundation through dmr-0846852 and the umass energy frontier research center through the department of energy . b. pansu , m .- h . li , and h.t . nguyen , j. phys . ii france * 7 * , 751 ( 1997 ) ; b. pansu , e. grelet , m .- h . li and h.t . nguyen , , 658 ( 2000).e . grelet , b. pansu , m .- h . li , and h.t . nguyen , , 3791 ( 2001 ) ; , 010703(r ) ( 2001 ) .	when thermal energies are weak , two dimensional lamellar structures confined on a curved substrate display complex patterns arising from the competition between layer bending and compression in the presence of geometric constraints . we present broad design principles to engineer the geometry of the underlying substrate so that a desired lamellar pattern can be obtained by self - assembly . two distinct physical effects are identified as key factors that contribute to the interaction between the shape of the underlying surface and the resulting lamellar morphology . the first is a local ordering field for the direction of each individual layer which tends to minimize its curvature with respect to the three - dimensional embedding . the second is a non - local effect controlled by the intrinsic geometry of the surface that forces the normals to the ( nearly incompressible ) layers to lie on geodesics , leading to caustic formation as in optics . as a result , different surface morphologies with predominantly positive or negative gaussian curvature can act as converging or diverging lenses respectively . by combining these ingredients , as one would with different optical elements , complex lamellar morphologies can be obtained . this smectic optometry enables the manipulation of lamellar configurations for the design of novel materials .
in general , many - body systems with interactions are very difficult to solve exactly . a way to overcome this difficulty is by approaching the many - body problem by a one - body problem , in which a mean - field replaces the interactions affecting the body . this idea is applied to the ferromagnetic ising model ( see reference @xcite ) . in the most simple mean - field approach , the nearest - neighbor interactions affecting each spin @xmath0 are replaced in such a way that @xmath0 now interacts with an effective field given by @xmath1 , where @xmath2 is the coordination number , @xmath3 the exchange constant , and @xmath4 is the thermal average of the spin @xmath5 . this is the so called `` weiss mean - field approach '' @xcite . nevertheless , it neglects the spin correlations , and it leads the transition temperature @xmath6 as well as the values of the critical exponents away from the exact values ( @xmath7 , for all dimensions ) . however , for the one - dimensional case , the ising model lacks of a phase transition at finite temperature , but weiss approach wrongly predicts that @xmath8 . a further step for improving the solution of this problem , is to use the proposal of hans bethe , which consists in considering that a central spin should interact with all its nearest - neighbor spins forming a cluster @xcite . then , that cluster would interact to an effective -field that approaches the next - nearest - neighbor spins surrounding the cluster . thus , this improvement gives @xmath9 , which not only betters the approximation of the critical temperatures , but leads correctly to @xmath10 , for the one - dimensional case . in this way , the correlations between the spins has been included to some degree by considering a cluster of spins interacting with its nearest - neighbors . the mean - field approach used in ising - like models with finite - size clusters is based on the following hamiltonian splitting : @xmath11 where @xmath12 corresponds to the energy that is composed of spin variables of the finite cluster , whereas @xmath13 , corresponds to the energy of the neighborhood , whose spins do not belong to the central sites of the finite cluster . in the canonical ensemble , the calculation of mean values of the spin variables @xmath14 belongs to the subspace @xmath15 of the finite cluster , and it is computed by the following procedure : @xmath16 this equation is exact if @xmath17 = 0 $ ] . the great merit of eq.([mo ] ) , is that we can solve the model of the infinite system by using a finite system in the subspace @xmath15 . various approximation methods use eq.([mo ] ) as a starting point . one of them is the effective - field theory ( eft ) proposed by honmura and kaneyoshi @xcite for solving the spin-1/2 ferromagnetic system . @xcite applied eft so as to treat different magnetic models with competing interactions . recently , viana _ et al . _ @xcite developed a mean - field proposal for spin models , denominated effective correlated mean - field ( ecmf ) , based on the following ansatz : @xmath18 where @xmath19 are the neighbors of the central spins of the finite cluster , @xmath20 is the mean of the spin variable of the cluster , and @xmath21 is a term exhibiting the behavior of a molecular parameter . there are many ways of determining the parameter @xmath22 . in the present work we use eq . ( [ rel3 ] ) by proposing that @xmath21 be the mean response of the spins of a monte carlo simulation ( smc ) , thus we make use of the following expression : @xmath23 where @xmath24 where @xmath25 is the number of monte carlo steps of the traditional metropolis algorithm , and @xmath26 is the total number of sites @xmath27 used in smc . in what follows , we call this new mean - field monte carlo technique as mft - mc . the aim of this proposal is the improvement of the results obtained by other mean - field techniques , but we believe that the main advantage of our mean - field proposal is the simplicity in which we can treat the first - order phase transitions . accordingly , in this work , we implemented this proposal in the spin-1 blume - capel model in the linear chain and in the square lattice . we remark that the blume - capel model ( @xmath28 ) @xcite is one of the most suitable models for studying magnetic systems from the point of view of the statistical mechanics . this model and its generalization , the blume - emery - griffiths model , ( @xmath29 ) was proposed to describe the @xmath30 transition in @xmath31 mixtures @xcite as well as ordering in a binary alloy @xcite . furthermore , its applications also include the description of ternary fluids @xcite , solid - liquid - gas mixtures and binary fluids @xcite , micro - emulsions @xcite , ordering in semiconducting alloys @xcite and electron conduction models @xcite . indeed , the bc model is found in many works using different lattices , spin degrees , including disorder and different statistical mechanic techniques @xcite . the bc model is represented by the following hamiltonian : @xmath32 where @xmath3 is the ferromagnetic coupling between the spins @xmath33 of the lattice and @xmath34 is the anisotropy parameter . when the value of @xmath34 increases from zero , the energy levels @xmath35 tend to the state @xmath36 , in such a way that when @xmath37 all the spins take the value @xmath36 and the system suffers a first - order phase transition , i.e. , the system goes from the ordered state ( @xmath38 ) to the state ( @xmath36 ) through a jump discontinuity . the critical value @xmath39 can be determined by equating the energy of the order and disordered states , i.e. , @xmath40 therefore , the bc model provides us a phase diagram with a a tricritical point separating a second and a first - order frontier that divides the ferromagnetic order ( f ) and the paramagnetic region ( pm ) . this rich critical behavior qualifies the bc model for representing different phase transitions , and our purpose is to apply the mft - mc method in it . the phase diagram of the bc model with equivalent - neighbor interactions can be seen in fig.1 of reference @xcite , which is qualitatively the same for dimensions greater than one . in order to study the ferromagnetic model in the square lattice , we use fig.1 as a reference . in this figure we may see that @xmath41 represents the neighbors of the central sites @xmath42 , which compose the finite cluster of @xmath43 sites . so , the hamiltonian of the bc model is conveniently written in the following form : @xmath44 where @xmath3 is the ferromagnetic coupling factor , @xmath45 is the number of spins @xmath41 interacting with @xmath27 and @xmath34 is the anisotropy parameter . note that eq . ( [ hf ] ) can be rewritten as follows : @xmath46 where @xmath47 ( @xmath48 ) , @xmath49 and @xmath50 in the present work the ansatz given in eq . ( [ rel3 ] ) is our fundamental assumption , so @xmath51 in this work we considered clusters containing @xmath52 central sites . thus , we have the following relations : @xmath53 the magnetization per particle @xmath54 of the magnetic system is given by the following definition : @xmath55 where the partition function in the space of sites @xmath42 is given by @xmath56 an important issue is the thermodynamic treatment of the spin system , accordingly , we use the free energy given by the following equation : @xmath57 where @xmath58 is the reduced temperature and @xmath59 is a parameter to be determined . at the equilibrium , the free energy is minimized , thus : @xmath60 where the function @xmath61 stands for the equation of state given by @xmath62 and we can determine the parameter @xmath59 by using eq . ( [ cond6 ] ) . we firstly obtained the phase diagram of the bc model in the plane @xmath63-@xmath64 in a square lattice , based on the considerations of the previous section . in fig . ( 2 ) we show the phase diagram for clusters containing @xmath65 ( in ( a ) ) and @xmath66 ( in ( b ) ) central sites . we faced the computational problem of solving the model for big clusters , inasmuch as the number of accessible states corresponds to @xmath67 states . for instance , for @xmath66 and @xmath68 sites , we have accessible states of order @xmath69 and @xmath70 , respectively , which are huge numbers . accordingly , for clusters of sizes @xmath71 , we prefer only to calculate the critical temperature @xmath72 , for @xmath73 , and the coordinates of the tricritical point @xmath74 . in order to plot the frontiers in fig . ( 2 ) , we used @xmath75 , for the size of the square lattice , and @xmath76 monte carlo steps in the simulation so as to obtain the molecular parameter @xmath21 . it should be noted that , for @xmath77 and @xmath78 , the results of the physical parameters obtained in the mean - field approach are just the same . in this figure we may observe that the critical value of the anisotropy corresponds to @xmath79 , for @xmath80 , which agrees with exact results obtained when equating the energy of the ordered state @xmath81 , with the energy of the disordered one ( @xmath82 ) , for a finite system of @xmath83 sites , i.e. , @xmath84 where @xmath85 is the coordination number of the lattice . the black circle represents the tricritical point that separates the second - order and the first - order frontier . we observe that the critical temperature @xmath72 decreases as the size of the cluster @xmath43 increases . the first - order frontier correctly falls perpendicularly to the anisotropy axis , however , in general , effective - field results do not reproduce this feature of the first - order frontier ( see the ieft curve in figure 5 of reference @xcite ) . in table 1 we present the values of @xmath72 , obtained for each cluster size @xmath43 , where we compare the results of this work using mft - mc with the mean - field approximation that uses clusters ( mft ) , where @xmath86 , in this case . we remark that yksel _ @xcite obtained @xmath87 , by using a metropolis monte carlo simulation , whereas silva _ @xcite obtained @xmath88 using wang - landau sampling . in references @xcite and @xcite we have @xmath89 and @xmath90 , respectively . we may observe that the results obtained by the mft - mc approach are close to the smc values when the cluster size is increased . nevertheless , if compared with the mft results , @xmath72 is better estimated by the mft - mc method , regarding the smc results as a reference . the calculations of the tricritical points @xmath74 through the mft - mc technique are shown in table 2 . we see that when the cluster size @xmath43 is increased , the values of the critical anisotropy @xmath91 approach the values @xmath92 and @xmath93 obtained by the monte carlo simulations of references @xcite and @xcite , respectively . in what the value @xmath94 respects , the tendency of the convergence is closer to that of reference @xcite , which gives @xmath95 , within their effective - field approach ( ieft ) . the coordinates of the tricritical point were determined through a landau expansion , @xmath96 from this equation , we are interested in solving the following system of equations @xmath97 in order to obtain the tricritical point . so , we may note that @xmath98 corresponds to the equation that determines the coefficients @xmath99 . in fig . ( 3 ) and fig . ( 4 ) we can see the behavior of the magnetization and the entropy , respectively . these are related to the phase diagram shown in fig . ( 2 ) , for @xmath66 . thus , the magnetization and the entropy curves , as functions of the temperature , were used to exemplify the first - order phase transition happening for @xmath100 , and the second - order phase transition , for @xmath101 . in fig . ( 3 ) we may observe that the magnetization falls down to zero continuously for the second - order phase transition ( see curve ( a ) ) , while it suffers of jump discontinuity for the first - order one ( see curve ( b ) ) . the study of the entropy per site is developed from the following expression : @xmath102 where @xmath103 means that all the other variables must be fixed in this differentiation . in fig . ( 4 ) the entropy exhibits an inflection point at the second - order transition ( in curve ( b ) ) , nevertheless , we may see a discontinuity at the first - order transition being characterized by @xmath104 ( see curve ( a ) ) , thus indicating the presence of a latent heat equal to @xmath105 , at the corresponding transition temperature . in the limit of high temperatures ( @xmath106 ) the entropy is @xmath107 , where the number @xmath108 is the number of accessible states for each spin of a spin-1 system . this result is correct in the canonical formalism . in order to study other lattices , we particularly consider the criticality of an uni - dimensional lattice . in fig . ( 5 ) we have the scheme of an uni - dimensional lattice used in modeling the mean - field theory . the terms @xmath109 correspond to the following expressions for each size @xmath43 : @xmath110 in table 3 we show the results of the critical temperature @xmath72 for the uni - dimensional lattice , running @xmath76 monte carlo steps so as to get the molecular parameter @xmath22 . we may observe that when the size of the cluster @xmath43 is increased the value of the critical temperature tends to zero , which is correct for an uni - dimensional system with nearest - neighbor interactions . so , though the convergence to the zero temperature is slow with @xmath43 , it proves that the mft - mc works in the lowest dimension . in this paper we study the ferromagnetic spin-1 blume - capel model with nearest - neighbor interactions , within a mean - field approach , which uses the magnetization response of a monte carlo simulation , as the molecular field . we call this new approach as mft - mc . for the bi - dimensional case , we implemented the model in the square lattice . the results show that when the size of the cluster @xmath43 increases , the values of the critical temperature @xmath72 ( for null anisotropy ) tend to @xmath111 , which is the monte carlo estimation of yksel _ et al . _ @xcite ( see table 1 ) . another important result is related to the estimation of the tricritical point @xmath74 in comparison with the results of silva @xcite and yksel@xcite ( see table 2 ) . our mft - mc values of @xmath91 reasonably agree with the results of the monte carlo simulations , as @xmath43 increases , whereas @xmath94 tends to the effective - field ( ieft ) result developed in reference @xcite . in order to test our mft - mc approach , we applied it to the uni - dimensional case , where we can verify through table 3 , that the critical temperature converges to zero for larger sizes . this shows that the mft - mc works well for uni - dimensional and bi - dimensional lattices . furthermore , this proposal allows an analysis of the thermodynamic properties after obtaining the entropy and the specific heat curves . on the other hand , the possibility of working with larger sizes constitutes an advantage for analyzing the finite - size effects . finally , we hope that this new approaching technique can be applied satisfactorily in other models and lattices .	we applied a mean - field approach associated to monte carlo simulations in order to study the spin-1 ferromagnetic blume - capel model in the square and the linear lattice . this new technique , which we call mft - mc , determines the molecular field as the magnetization response of a monte carlo simulation . the resulting phase diagram is qualitatively correct , in contrast to effective - field approximations , in which the first - order line is not perpendicular to the anisotropy axis at low temperatures . thermodynamic quantities , as the entropy and the specific heat curves can be obtained so as to analyze the nature of the phase transition points . also , the possibility of using larger sizes constitutes an improvement regarding other mean - field approximations that use clusters . * pacs numbers * : 64.60.ak ; 64.60.fr ; 68.35.rh
in this study we consider the problem of parameter - free identification of a class of dynamical systems . the approach we propose is derived from a general method for the reconstruction of the constitutive relations in systems described by partial differential equations ( pdes ) which was initially introduced in @xcite and further developed in @xcite , see also @xcite the idea is that , given an autonomous evolution equation @xmath0 for some quantity @xmath1 ( defined in a finite or infinite dimension with @xmath2 denoting time ) , one seeks to optimally reconstruct the flux function @xmath3 , so that the system outputs best match , in a suitably defined sense , with the measurements available . as shown schematically in figure [ fig : f ] , the originality of this approach is that the function @xmath3 is reconstructed directly as a _ continuous _ object , rather than employing a truncated polynomial or other galerkin expansion . in @xcite we reviewed the mathematical foundations and some computational aspects of this approach applied to the reconstruction of state - dependent transport coefficients in a class of systems described by pdes . in the present investigation we adapt this method to the problem of identification of the propagator function @xmath4 , @xmath5^t$ ] , in a finite - dimensional dynamical system @xmath6 with the state vector @xmath7^t \in \rr^n$ ] . to fix attention , instead of working with an abstract formulation , we will focus on a specific problem concerning a three - dimensional ( @xmath8 ) model characterized by oscillatory dynamics with a stable limit cycle . in many real - life physical problems with infinite - dimensional state spaces and evolution described by pdes such systems arise as reduced - order models providing low - dimensional approximate description in the neighborhood of fixed points and limit cycles . in some cases propagator @xmath9 may be derived from the description of the full plant based on first principles . in general , however , the reduced - order model is inferred from experimental or numerical data . such reduced models are of paramount importance as test - beds for the development of our understanding of system dynamics . they are also useful as low - cost surrogates guiding optimization and real - time control design for expensive full - scale models . in all cases , typical model identification is generally performed in three steps : ( 1 ) selection of the state space @xmath10 which is large enough to capture the behaviour of interest and at the same time sufficiently small to allow one to exploit the analytical / numerical advantages of the surrogate plant ; ( 2 ) structure identification of the propagator @xmath11 , e.g. , determination of the polynomial degree of its components @xmath12 , @xmath13 ; and ( 3 ) parameter identification , e.g. , inference of the polynomial coefficients from time - resolved trajectories @xmath14 . these different steps are illustrated schematically in figure [ fig : f]a . the challenge of this approach is to find the right balance between the robustness of the model identification , requiring only a few tunable parameters , and a good accuracy for which a larger number of parameters is typically needed . identification problems can also be solved using variational techniques and this is the approach we will pursue in the present study , cf . figure [ fig : f]b . similar methods have been developed for a broad range of problems in both the finite and infinite - dimensional setting , including flow control in fluid mechanics @xcite , data assimilation in meteorology @xcite and geophysics @xcite to mention just a few application areas . the related problem of state estimation is usually solved using various filtering approaches such as the kalman filter @xcite . we introduce now our model . the oscillatory fluctuation is parameterized by @xmath15 where @xmath16 is the imaginary unit , @xmath17 is the amplitude of the fluctuation ( `` @xmath18 '' means `` equal to by definition '' ) and @xmath19 the corresponding phase , while the base - flow deformation is characterized by a single parameter @xmath20 . following the mean - field theory @xcite , we make the following assumptions about the structure of the system : 1 . in the plane @xmath21 the system exhibits an unstable fixed point at the origin and an attracting limit cycle , 2 . the state variable @xmath20 is `` slaved '' to @xmath22 and @xmath23 , i.e. , @xmath24 , and 3 . the dynamics is phase - invariant , i.e. , the descriptor system depends only on @xmath25 . [ ass1 ] as regards the time @xmath26 , we will assume that @xmath27 $ ] for some @xmath28 . as a general form of a dynamical system consistent with assumption [ ass1 ] we will consider [ eq : descriptorsystemb ] @xmath29 or , equivalently [ eq : descriptorsystema ] @xmath30 where @xmath31 and @xmath32 . subsequently , we will also use the notation @xmath33^t$ ] and @xmath34 $ ] , where @xmath35 } r(t)$ ] . we note that the equations governing @xmath22 and @xmath23 , or equivalently @xmath25 and @xmath36 , do not depend on @xmath20 . equations and describe the dependency of a slow variable on the fluctuation amplitude , and the usefulness of these algebraic equations will become clear in the context of the mean - field models discussed in section [ toc : meanfieldsystem ] . in and the functions @xmath37 , @xmath38 , are assumed sufficiently regular to make the systems well posed ( the question of the regularity of functions @xmath39 , @xmath38 , will play an important role in our approach and will be addressed in detail further below ) . without loss of generality , we will also assume that @xmath40 . system , or , is complemented with the initial condition , respectively , @xmath41^t$ ] and @xmath42 , @xmath43 . dynamical systems of the type or arise commonly as a result of various rigorous and empirical model - reduction strategies in diverse application areas such as fluid mechanics @xcite , thermodynamics @xcite and phase transitions @xcite . the main contribution of this work is development of a computational technique allowing one to reconstruct functions @xmath44 , @xmath45 and @xmath46 in in a very general form based on some measurements . the key idea is to formulate a least - squares minimization problem in which one of the functions @xmath39 , @xmath38 , is the control variable . then , a variational gradient - based approach can be employed to find the optimal solution in a suitable function space . in the present investigation , we will consider system as a reduced - order model of hydrodynamic instabilities in open shear flows obtained using a suitable galerkin projection to fix attention without losing generality . although this problem is very well - studied , our method provides a systematic refinement of the state - of - the - art mean - field model , which is the second main contribution of this study . details of this problem are introduced in the next section . in the following section we develop our model identification approach , whereas in section [ toc : results ] we present a number of computational results concerning identification of the model for the system considered in section [ toc : modelidentification ] . then , in section [ toc : discussion ] , we analyze the computational performance of our method and discuss the improvements it offers over the predictions of some standard models applied to the problem in question . summary and conclusions are deferred to section [ toc : conclusions ] , whereas in appendices [ toc : pod ] , [ toc : regularity ] and [ toc : perturbation ] we collect some technical results . in this section we define a model identification problem associated with the transient two - dimensional ( 2d ) cylinder wake which is a well - studied hydrodynamic instability @xcite . this flow is a representative example of phenomena characterized by the hopf bifurcation with an unstable fixed point and a stable limit cycle . additional examples in this category include the rossiter modes of the flow over a cavity @xcite and other shear flows @xcite . first , in section [ toc : wake ] , we describe the initial - boundary value problem for the infinite - dimensional navier - stokes equation which is a system of coupled pdes representing the conservation of mass and momentum in the motion of viscous incompressible fluid . then , in section [ toc : galerkinexpansion ] , a low - dimensional galerkin expansion is recalled which reduces the kinematic description down to three modes whose amplitudes serve as the state variables @xmath22 , @xmath23 and @xmath20 . it will be demonstrated that the system governing these variables satisfies in fact assumptions [ ass1 ] and is in the form . special forms of this system arising as models in numerous applications are discussed in section [ toc : meanfieldsystem ] , whereas in section [ toc : measurements ] we describe the measurements used as the basis for the reconstructions . the 2d flow around a circular cylinder is described here in the cartesian coordinate system @xmath47 . the origin @xmath48 coincides with the center of the cylinder , the @xmath49-coordinate points in streamwise direction , while @xmath50 represents the transverse coordinate . the velocity field @xmath51 has components @xmath52 and @xmath53 aligned with the @xmath49-axis and @xmath50-axis , respectively , whereas @xmath54 represents the pressure field . the oncoming flow velocity is denoted by @xmath55 and the cylinder diameter by @xmath56 . the newtonian fluid is characterized by uniform density @xmath57 and kinematic viscosity @xmath58 . the flow properties depend on the reynolds number @xmath59 . here , we consider the case of @xmath60 which is far above the critical reynolds number of @xmath61 characterizing the onset of vortex shedding ( i.e. , the hopf bifurcation ) @xcite , and far below the transitional reynolds number of 187 which marks the onset of three - dimensional instabilities @xcite . in the following , all quantities are assumed to be non - dimensionalized with @xmath55 , @xmath56 and @xmath57 . the flow is considered in a rectangular domain surrounding the cylinder @xmath62 and its evolution is described by the incompressible navier - stokes equation [ eqn : incompressibleflow ] @xmath63 the boundary conditions for the velocity comprise the no - slip condition at the cylinder , the free - stream condition at the inlet boundary , the free - slip condition at the lateral boundaries and the no - stress condition at the downstream outflow boundary . the initial condition for the velocity at @xmath64 is the unstable fixed point @xmath65 of navier - stokes equation perturbed by the real part of the corresponding most unstable stability eigenmode @xmath66 @xmath67 the field @xmath66 is normalized to have unit @xmath68 norm . the initial - boundary - value problem is numerically integrated with a finite - element method based on an unstructured grid and details are provided in @xcite . figure [ fig : flows ] depicts three flow snapshots corresponding to the initial condition , an intermediate transient state and the flow approaching the limit cycle . [ fig : flows ] \(a ) + ( initial condition ) , ( b ) @xmath69 ( intermediate transient state ) , and ( c ) @xmath70 ( periodic solution at the limit cycle ) . the domain shown represents the computational domain @xmath71 which also coincides with the region where the galerkin expansion is defined . the flow patterns are visualized using streamlines ( which are defined as the level sets of the streamfunction @xmath72 related to the velocity components via @xmath73 and @xmath74 ) . the cylinder is indicated by the black circle . , title="fig:",width=264 ] + ( b ) + ( initial condition ) , ( b ) @xmath69 ( intermediate transient state ) , and ( c ) @xmath70 ( periodic solution at the limit cycle ) . the domain shown represents the computational domain @xmath71 which also coincides with the region where the galerkin expansion is defined . the flow patterns are visualized using streamlines ( which are defined as the level sets of the streamfunction @xmath72 related to the velocity components via @xmath73 and @xmath74 ) . the cylinder is indicated by the black circle . , title="fig:",width=264 ] + ( c ) + ( initial condition ) , ( b ) @xmath69 ( intermediate transient state ) , and ( c ) @xmath70 ( periodic solution at the limit cycle ) . the domain shown represents the computational domain @xmath71 which also coincides with the region where the galerkin expansion is defined . the flow patterns are visualized using streamlines ( which are defined as the level sets of the streamfunction @xmath72 related to the velocity components via @xmath73 and @xmath74 ) . the cylinder is indicated by the black circle . , title="fig:",width=264 ] below we review approximate modelling approaches typically employed to obtain low - dimensional descriptions of the dynamics described by , cf . @xcite . the quantities corresponding to the limit cycle will be denoted with the superscript @xmath75 and in the present problem in which the limit cycle is stable we will therefore have @xmath76 . the transient wake is usually characterized by the superposition of a time - varying symmetric base flow @xmath77 with the length of the recirculating region decreasing with time as the vortex shedding develops and an antisymmetric oscillating field @xmath78 representing the vortex shedding . the change of the base flow is accurately captured by a single `` shift mode '' @xmath79 @xcite , while the oscillation field can be approximated by the first two proper orthogonal decomposition ( pod ) modes @xmath80 , @xmath81 , computed at the limit cycle @xcite . hereafter , the subscript @xmath82 will denote various quantities related to the shift mode @xmath79 . some information about the construction of orthogonal bases using the pod approach is presented in appendix [ toc : pod ] . the resulting truncated galerkin expansion thus takes the form [ eqn : meanfieldexpansion ] @xmath83 one typically ignores deformations of the oscillatory modes during the transient . these deformations do not significantly alter the model predictions and do not have any effect on the proposed model identification approach . on the other hand , inclusion of this effect in our model would significantly complicate the study and is outside the scope of this investigation . identifying @xmath84 and assuming phase - invariant behavior , we note that the evolution of the galerkin expansion coefficients is governed by a system in the form , provided that @xmath85 , corresponding to an unstable fixed point at the origin , and @xmath86 and @xmath87 , corresponding to a locally attracting limit cycle at @xmath88 , cf . assumption [ ass1](a ) . moreover , from the assumptions made in section [ toc : introduction ] , it follows that @xmath89 . introducing a number of further simplifications one arrives at two well - known reduced models , the mean - field model and the landau model . since they provide a point of reference for our test problems , we describe them briefly below . the mean - field model of oscillatory flow instabilities @xcite explicates the mechanism of amplitude saturation in the mean - field deformation . this deformation is quantified by the amplitude @xmath90 of the shift mode , cf . , which is slaved to the fluctuation level @xmath25 . for the soft ( supercritical ) hopf bifurcation , the resulting evolution equations read [ eqn : meanfieldsystem ] @xmath91 \ > r{(t ) } , \\ \label{eqn : meanfieldsystem_2 } \dot \theta{(t ) } & = & \omega_1 + \gamma_{\delta } a_{\delta}({t } ) , \\ \label{eqn : meanfieldsystem_3 } a_{\delta}({t } ) & = & \alpha_{\delta } \ > r^2{(t)}.\end{aligned}\ ] ] the ordinary differential equations correspond to the navier - stokes equation linearized around the time - varying base flow with @xmath90 as the order parameter . the third algebraic equation represents the reynolds equation linking the mean flow to the reynolds stress generated by the fluctuations described by the first equation . evidently , equations are a particular case of with @xmath92 , @xmath93 and @xmath94 . the parameters @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 of may be derived from the galerkin approximation described in section [ toc : galerkinexpansion ] , or identified from the solutions of the navier - stokes equation via suitable fitting . the periodic solution of is given by [ eqn : meanfieldsystemsolution ] @xmath100 at the limit cycle the mean flow is predicted to have a vanishing growth rate @xmath44 . this marginal stability property of mean flows has been conjectured by malkus @xcite and is corroborated by the global stability analysis of the navier - stokes equation @xcite . derivation details and a further discussion of the properties of the mean - field model can be found in @xcite . the landau equation for the supercritical hopf bifurcation is a corollary to the mean - field model and constitutes a prototype evolution equation for self - amplified , amplitude - limited oscillations . it is obtained by substituting in : [ eqn : landauequation ] @xmath101 here , we require @xmath102 to ensure a stable limit cycle with a positive angular velocity in the @xmath21 plane , cf . assumption [ ass1](a ) , whereas @xmath103 may vanish , be positive or negative . evidently , equations correspond to with @xmath104 and @xmath105 . therefore , the landau equation is also easily recognized as a particular case of and , while the mean - field equation is an example of and . the dependency of the growth rate and frequency on the shift mode amplitude @xmath90 in may in principle be recovered from by inverting to give @xmath106 and substituting @xmath107 into . this inversion assumes a monotonous dependence of @xmath90 on @xmath25 which is observed in actual wake data . for completeness , we note that the least - order galerkin model introduced in @xcite includes a fast - dynamics equation for @xmath90 . this dynamic equation is well represented by an inertial manifold @xmath108 of the form or which ignores short transients . the landau equation may also be derived from the first principles , e.g. , using a center - manifold reduction method , see , e.g. , @xcite . the parameters may be identified from solutions of the navier - stokes equation with @xmath95 and @xmath96 obtained as the real and imaginary part of the most unstable eigenvalue associated with the hopf bifurcation , whereas the nonlinearity parameters @xmath109 and @xmath103 can be inferred from the post - transient amplitude and frequency , @xmath110 and @xmath111 , respectively . in section [ toc : results ] we will demonstrate using our proposed approach how the structure of landau model could be modified to better reproduce the actual behavior . the state in our approximate model is characterized by three time - dependent mode amplitudes : @xmath22 , @xmath23 and @xmath20 and as `` measurements '' we will consider the functions @xmath112 , @xmath113 and @xmath114 which are obtained for @xmath115 $ ] by solving initial - boundary - value problem - followed by projection , in terms of the inner product defining the pod analysis ( cf . appendix [ toc : pod ] ) , of the resulting time - dependent velocity field @xmath116 on the modes @xmath117 , @xmath118 and @xmath79 . the same procedure applies when the velocity field comes from time- and space - resolved measurements . in either case , determination of the shift mode @xmath79 requires access to the unstable equilibrium @xmath65 , cf . , which typically needs to be obtained numerically . the mean - field model may also be constructed directly from experimental measurements . let @xmath119 be , for instance , a pressure or hot - wire signal with a dominant harmonic component corresponding to the vortex shedding . let @xmath120 be the short - time mean value , e.g. , a one - period average , and @xmath121 be the fluctuation . next , let @xmath22 and @xmath23 be the local cosine and sine component of the fluctuation . these variables may be obtained from a hilbert transform or , more robustly , from a morlet transform of the data . finally , we identify @xmath122 , where the tunable constant shall ensure the correct fixed - point behavior , i.e. , @xmath123 when @xmath124 . then , @xmath22 , @xmath23 and @xmath90 may be approximated by the mean - field model . in this case , it is not required that the fixed point be actually reached for the identification of the model . we only need a transient with a range of @xmath25 values over which the functions @xmath39 are identified . in the next section we introduce a computational approach allowing one to optimally identify the functions @xmath125 , @xmath126 and @xmath127 in a suitable class , so that the predictions of system best match in the least - squares sense the data @xmath128 , @xmath129 and @xmath130 . we note that , alternatively , a system of evolution equations for @xmath22 , @xmath23 and @xmath20 may be obtained by substituting ansatz into momentum equation , and some comparisons between such an approach and the results obtained using the model identification method developed here will be drawn in section [ toc : discussion_physical ] . the task of identifying the functions @xmath39 , @xmath38 , such that the output of system matches certain `` measurements '' is an example of an inverse problem @xcite . what makes this problem somewhat different from typical inverse problems is that the functions sought have the form of `` constitutive relations '' , in the sense that they depend on the _ state variables _ ( i.e. , the dependent variables in the problem ) , rather than the independent variables . more specifically , in the problem considered here @xmath39 , @xmath38 , depend on @xmath131 as opposed to @xmath26 . non - parametric formulations of such inverse problems have received some attention in the context of systems described by pdes @xcite , but we are not aware of similar approaches applied to the state - space description of dynamical systems . we will look for functions @xmath39 , @xmath38 , as elements of the sobolev space @xmath132 of continuous functions with square - integrable gradients on @xmath133 which is the `` identifiability '' region defined in section [ toc : introduction ] , i.e. , the interval spanned by the state variable @xmath134 during the system evolution , see also @xcite . some additional remarks concerning the regularity of functions @xmath44 and @xmath45 are presented in appendix [ toc : regularity ] . in the reconstruction problems considered in this study , the boundary behavior of the reconstructed functions will have to be restricted in different ways . this reflects the fact that system with the reconstructed functions @xmath39 , @xmath38 , should exhibit the behavior dictated by assumption [ ass1](a ) in specific regions of the phase space , namely , at the equilibrium @xmath124 and at the limit cycle @xmath135 . there is some flexibility as regards possible choices and the boundary conditions we adopt will reproduce the behavior described by the mean - field model introduced in section [ toc : meanfieldsystem ] . more specifically , * at the origin @xmath124 * * the jacobian of the right - hand side ( rhs ) in equation should be given by @xmath136 which is a priori unknown and is to be determined as a part of the solution of the reconstruction problem ; on the other hand , the jacobian of the rhs in equation should vanish ; this is achieved when the derivatives of both functions @xmath44 and @xmath45 are set to zero at the origin , i.e. , @xmath137 * at the limit cycle @xmath138 * * the rhs of should vanish resulting in the vanishing of @xmath44 , i.e. , @xmath139 * * in regard to equation , we will prescribe a given slope @xmath140 of the rhs , i.e. , @xmath141 we note that , while the boundary behavior described by at the equilibrium and the limit cycle is the same as in the mean - field model ( cf . section [ toc : meanfieldsystem ] ) , the behavior of the constitutive relations @xmath44 and @xmath45 for intermediate values of the state magnitude @xmath142 can be arbitrary and will be determined using our optimal reconstruction procedure . no restrictions are placed on the boundary behavior of function @xmath46 , cf . . a convenient way to solve inverse problems is to formulate them as suitable optimization problems @xcite . for each of the functions @xmath39 , @xmath38 , we thus define the corresponding cost functional @xmath143 as [ eq : j ] @xmath144 ^ 2 \ , dt , \label{eq : j1 } \\ \j_2(g_2 ) & : = \frac{1}{2 } \int_0^t \left [ e^{\imath \theta(t ) } - e^{\imath \tilde{\theta}(t ) } \right]^2 \ , dt , \label{eq : j2 } \\ \j_3(g_3 ) & : = \frac{1}{2 } \int_0^t \left [ g_3(r(t ) ) - \ta_{\delta}(t ) \right]^2 \ , dt , \label{eq : j3 } \end{aligned}\ ] ] where @xmath145 , @xmath146 and @xmath114 are the `` measurements '' obtained as described in section [ toc : measurements ] . the length @xmath147 of the assimilation window will be chosen sufficiently long to allow the transient trajectory to settle on the limit cycle . in and the functions @xmath134 and @xmath148 are related to @xmath44 and @xmath45 via system . the optimal reconstructions @xmath149 , @xmath150 , and @xmath151 are defined as solutions of the following problems [ eq : p ] @xmath152 below we describe in detail a gradient - based approach to solution of problem @xmath153 . problem @xmath154 has a similar structure and the solution method is essentially the same with some modifications described hereafter . in problems @xmath155 and @xmath156 it is assumed that the functions , respectively , @xmath45 and @xmath44 are fixed . problem @xmath157 has a different structure warranting a separate solution approach which will be described further below . the minimizer @xmath149 is characterized by the first - order optimality condition @xcite requiring the vanishing of the gteaux differential + @xmath158 $ ] , i.e. , @xmath159 where @xmath160 is an arbitrary perturbation direction . the ( local ) minimizer can be computed with the following iterative procedure @xmath161 where @xmath162 represents the initial guess , @xmath163 denotes the iteration count and @xmath164 is the gradient of cost functional @xmath165 . the length @xmath166 of the step is determined by solving the following line minimization problem @xmath167 which can be done efficiently using standard techniques such as brent s method @xcite . for the sake of clarity , formulation represents the steepest - descent method , however , in practice one typically uses more advanced minimization techniques , such as the conjugate gradient method , or one of the quasi - newton techniques @xcite . evidently , the key element of minimization algorithm is the computation of the cost functional gradient @xmath168 . it ought to be emphasized that , while the governing system is finite - dimensional , the gradient @xmath168 is a function of the state magnitude @xmath25 and as such represents a continuous ( infinite - dimensional ) sensitivity of cost functional @xmath169 to the perturbations @xmath170 . the fact that the control variable @xmath44 , and hence also the gradient @xmath171 , are functions of the _ state _ variable , rather than the independent variable ( figure [ fig : vars]a ) , will result in cost functional gradients with structure rather different than encountered in typical optimization problems for differential equations ( see @xcite for some related questions arising in pde optimization problems ) . in order to identify an expression for the gradient @xmath171 , we proceed by computing the gteaux differential of the cost functional @xmath172 r'(g_1;g'_1)\ , dt = \int_0^t \frac{r - \tr}{r } \bxi^t\bxi ' \ , dt = \big\langle \nabla\j_1(g_1 ) , g'_1 \big\rangle_{\x(\i ) } , \label{eq : dj}\ ] ] where we used the identity @xmath173 and the last equality is a consequence of the riesz representation theorem @xcite with @xmath174 denoting an inner product in the hilbert space @xmath175 ( to be specified later ) of functions defined on @xmath133 . the perturbation variable @xmath176 is a solution of the following perturbation problem ( see appendix [ toc : perturbation ] for a derivation ) [ eq : dxi ] @xmath177 \bxi'(t ) + \bi \ , \bxi(t)\ , g'_1 \nonumber \\ & = : \ba(\bxi(t ) ) \ , \bxi ' + \bi \ , \bxi(t)\ , g'_1 , \label{eq : dxi1 } \\ \bxi'(0 ) & = 0 , \label{eq : dxi2}\end{aligned}\ ] ] where @xmath178^t$ ] , @xmath81 . we note that gteaux differential is not yet in the form consistent with the riesz representation , since the perturbation @xmath160 does not appear in it as a factor , but is hidden on the rhs in perturbation equation . a standard technique to convert gteaux differential to the riesz form is based on the _ adjoint _ variable @xmath179\rightarrow \rr^2 $ ] . taking the inner product ( in @xmath180 ) of @xmath181 with equation , integrating over @xmath182 $ ] and then integrating by parts we obtain @xmath183 \bxi ' - \bxi\ , g'_1\big\}\ , dt \\ = & \int_0^t \ , \left ( \bxi'\right)^t \big\{-\dot{\bxi}^ * - \left [ g_1(r ) \bi + \bi\ , \bxi\ , ( \bnabla g_1(r))^t + g_2(r ) \bj + \bj\ , \bxi\ , ( \bnabla g_2(r))^t \right]^t \bxi^ * \big\ } \ , dt \\ & + \big [ ( \bxi^)^t \bxi'\big]_{t=0}^{t = t } - \int_0^t \ , \left ( \bxi^\right)^t \ , \bi \ , \bxi(t)\ , g'_1 \ , dt . \end{split } \label{eq : id1}\ ] ] defining the _ adjoint _ system as [ eq : axi ] @xmath184^t \bxi^(t ) + \frac{r - \tr}{r } \bxi , \label{eq : axi_1 } \\ \bxi^(t ) & = 0 , \label{eq : axi_2}\end{aligned}\ ] ] we reduce relation to @xmath185 we note that , although @xmath160 already appears as a factor in expression , this expression is still not in the riesz form , since the integration is with respect to the time @xmath186 , whereas in the inner product @xmath174 defining the riesz representer integration is with respect to the measure @xmath187 defined on the interval @xmath133 ( connection between the different integration variables is illustrated schematically in figure [ fig : vars]b ) . the two variables are related via the following transformation @xmath188 where we used the identities @xmath189 and @xmath190 , cf . . denoting the trajectory in the state space @xmath191 } \ \bxi(t ) \in \rr^2 \right\}$ ] , and combining with we obtain the expression @xmath192 which is already in the required riesz form . in the line integral over the contour @xmath193 and the definite integral over the interval @xmath133 are equal , because for dynamical system points on the contour @xmath193 with the magnitude @xmath194 are unique , so that the map @xmath195 is one - to - one ( in more general situations when this is not the case , or when the reconstructed function depends on more than one state variable , e.g. , both @xmath196 and @xmath197 here , the change of variables needed to obtain the riesz form will be more complicated and one has to employ more general techniques such as those developed in @xcite ) . while this is not the gradient we will use in actual computations , we will first obtain an expression for the @xmath198 gradient which in the next subsection will be used as the basis for the calculation of gradients defined in the sobolev space @xmath132 . thus , setting @xmath199 in , we obtain from @xmath200 expression is validated computationally in section [ toc : validation ] . as regards problem @xmath156 , the optimality condition takes the form , cf . and , @xmath201 where we used the identity @xmath202 . following the same steps as described above , we obtain an expression for the cost functional gradient in the form , however , the adjoint system satisfied by @xmath203 has now a different source term on the rhs [ eq : axi2 ] @xmath184^t \bxi^(t ) + \frac{\sin(\theta - \tilde{\theta})}{r^2 } \bj\ , \bxi(t ) , \label{eq : axi2_1 } \\ \bxi^(t ) & = 0 . \label{eq : axi2_2}\end{aligned}\ ] ] in regard to problem @xmath204 , using change of variables , we can rewrite as @xmath205 ^ 2 \ , dr . \label{eq : j3b}\ ] ] thus , in optimization problem we look for a function @xmath206 which is as close as possible ( in a weighted @xmath198 topology ) to a given function @xmath207 , this problem , in fact , does not have a solution because of the density of the function space @xmath132 in @xmath208 , cf . however , it is possible ( and satisfactory from the application point of view ) to `` solve '' problem approximately by finding a @xmath209 such that @xmath210 is sufficiently small . such an approach is described in section [ toc : sobolev ] . in this section we describe how sobolev gradients @xmath211 , @xmath81 , used in minimization algorithm for problems @xmath155 and @xmath156 can be obtained from . in addition to enforcing smoothness of the reconstructed functions , this formulation allows us to impose the desired behavior at the endpoints of the interval @xmath133 , cf . - , via suitable boundary conditions . we begin by defining the @xmath212 inner product on @xmath133 as @xmath213 where @xmath214 is a parameter with the meaning of a `` length scale '' . it is well known @xcite that extraction of cost functional gradients in the space @xmath212 with the inner product defined as in can be regarded as low - pass filtering the @xmath198 gradients with the cut - off wavenumber given by @xmath215 . as regards the behavior of the gradients @xmath216 at the endpoints of the interval @xmath133 , we can require the vanishing of either the gradient itself or its derivative @xmath217 , and the boundary conditions we prescribe correspond to relations introduced as a part of the formulation of optimization problems , cf . assumption [ ass1](a ) . as regards the boundary data at @xmath135 ( i.e. , at the limit cycle ) , in problem @xmath156 we use @xmath218 which ensures that the property @xmath219 of the initial guess @xmath220 remains unchanged during iterations . identifying expression with inner product , cf . , integrating by parts and using the boundary conditions mentioned above we obtain the following elliptic boundary - value problem on @xmath133 defining the sobolev gradient @xmath216 [ eq : gradjh1 ] @xmath221 where the expression for @xmath222 is given in . as concerns problem @xmath204 , we propose to reconstruct @xmath209 directly ( i.e. , without iterations ) by solving the following problem [ eq : gradj3 ] @xmath223 which , except for the boundary condition at @xmath138 , has an identical structure as . the superscript in @xmath224 indicates dependence of the solution on the parameter @xmath225 . we note that as @xmath226 the left - hand side ( lhs ) in approaches the identity transformation which means that @xmath227 , so that also @xmath228 , as @xmath226 . since solutions of system are not defined for @xmath229 , we will obtain our approximate reconstruction as @xmath230 for some small value of @xmath225 . results concerning model identification for the system described in section [ toc : modelidentification ] are presented in section [ toc : results ] , whereas in sections [ toc : validation ] and [ toc : discussion_computational ] we analyze certain computational aspects of the method . in this section we present results concerning the solution of model identification problems @xmath155 , @xmath156 and @xmath204 , cf . , for the system introduced in section [ toc : modelidentification ] . motivated by practical considerations , we make the following in the solution of problems @xmath155 and @xmath156 we set in system [ eq : g1g2 ] @xmath231 [ ass2 ] which means that in the reconstruction of @xmath45 we use the best available estimate of @xmath44 obtained from the solution of problem @xmath155 . the choice of @xmath45 has no effect on the solution of problem @xmath155 and hence without loss of generality we can adopt . as the gradient descent algorithm in problems @xmath155 and @xmath156 we use the polak - ribiere version of the nonlinear conjugate gradient method @xcite in which the `` momentum '' term is reset to zero every 20 iterations and iterations are declared converged when @xmath232 , @xmath81 . in the system described in section [ toc : modelidentification ] the limit cycle is characterized by @xmath233 , whereas the length of the time window is chosen as @xmath234 which is long enough to allow the transient to settle on the limit cycle ( see figure [ fig : ra1a3]a below ) . we have successfully solved problems @xmath155 , @xmath156 and @xmath204 using different combinations of numerical parameters , and the parameters used to obtain the results presented in this section are summarized below . systems and were solved using matlab subroutine ode45 with an adaptive time - stepping . unless stated otherwise , the integrals defined on the interval @xmath182 $ ] , cf . , were discretized using @xmath235 equispaced points . the interval @xmath133 was discretized using @xmath236 equispaced points , and boundary - value problems and were approximated using the second - order finite differences . the length - scale parameter appearing in and was @xmath237 in problems @xmath155 and @xmath156 , and @xmath238 in problem @xmath204 . the initial condition @xmath239 for system and the initial guesses @xmath162 and @xmath220 for reconstruction algorithm must be chosen so that the magnitudes @xmath240 , @xmath115 $ ] , span the entire interval @xmath133 , as otherwise the sensitivities ( gradients ) can not be properly defined for _ all _ values of @xmath25 ( we refer the reader to @xcite for a discussion how this limitation can be overcome in some cases ) . since our goal is now to assess possible improvements to mean - field model , we will use it with the coefficients determined as discussed in section [ toc : meanfieldsystem ] as the initial guess for the reconstructions , so that @xmath241 respectively , for problems @xmath155 and @xmath156 . it is clear that initial guesses satisfy properties with @xmath242 . as the initial condition @xmath239 for we used a small perturbation around the fixed point at the origin . our main results are presented in figure [ fig : g123 ] where we show the optimal reconstructions @xmath243 , @xmath38 , and compare them against the left - hand sides of equations , all shown as functions of the state magnitude @xmath25 . for completeness , the lhs of equations are shown as functions of time @xmath27 $ ] in figure [ fig : g12_vs_t ] ( in the case of @xmath149 , cf . figures [ fig : g123]a and [ fig : g12_vs_t ] , the lhs of equation is additionally divided into @xmath25 ) . in figures [ fig : g123]a , b we also indicate the rhs of mean - field model which were used as the initial guesses for the reconstructions . we see in figures [ fig : g123 ] that , as expected , the reconstructed constitutive relations @xmath243 , @xmath38 , smoothly approximate the left - hand sides of the corresponding equations evaluated using the measurements . considered as functions of @xmath25 , these left - hand sides are multi - valued which is a consequence of the fact that , due to the oscillations of the measurement data at the limit cycle ( see figure [ fig : ra1a3]a below ) , the map @xmath244 is not one - to - one . in figures [ fig : g123]a , b we observe systematic deviations of the optimal reconstructions @xmath149 and @xmath150 from the corresponding functions in mean - field model . in addition , based on the reconstructions @xmath149 and @xmath150 we can obtain estimates of two important quantities , namely , the growth rate of the instability at the origin given by @xmath245_{r=0 } = \hat{g}_1(0 ) = 0.1576 $ ] , and the oscillation frequency at the limit cycle given by @xmath246 . these numbers should be compared with , respectively , 0.151 and 1.036 obtained as discussed in section [ toc : meanfieldsystem ] and used in mean - field model . finally , in figure [ fig : ra1a3 ] we compare the outputs from system obtained using the mean - field model and the reconstructions @xmath243 , @xmath38 , against the corresponding measured quantities ( as regards the time - history of the state variables , we do not show @xmath247 , as it has qualitatively very similar behavior to @xmath248 already shown in figure [ fig : ra1a3]b ) . in figures [ fig : ra1a3]a , b ( see , in particular , the insets ) we note that the evolution of @xmath134 and @xmath248 obtained using the optimal reconstructions @xmath149 and @xmath150 is much closer to the measured quantities than the evolutions computed using mean - field model . we remark , however , that the measurements @xmath249 shown in figure [ fig : ra1a3]a reveal some high - frequency oscillations which are not captured by the trajectory @xmath134 obtained using the optimal reconstruction @xmath149 . these oscillations reflect a phase dependence in the behavior of the solutions of the original navier - stokes equation , an effect which is by construction excluded from ansatz , cf . assumption [ ass1](a ) . as a consequence , @xmath149 can convey phase - averaged information only . we will return to this problem again in section [ toc : discussion_physical ] . as regards the results shown in figure [ fig : ra1a3]c , we see that , while the optimal reconstruction @xmath250 is quite smooth ( figure [ fig : g123]c ) , the quantity @xmath251 exhibits oscillations absent in the original measurement data @xmath252 . this effect as well is a consequence of the lack of phase - dependence in ansatz and . in this section we analyze a number of computational and physical modelling aspects of the proposed approach which can be important in applications . we begin by examining the accuracy of the cost functional gradients in section [ toc : validation ] , followed by a study of the robustness of iterations in section [ toc : discussion_computational ] and conclude with some insights about physical modelling in section [ toc : discussion_physical ] . a key element of optimization algorithm are the cost functional gradients and a standard approach to their validation consists in computing the directional gteaux differential @xmath253 , @xmath81 , for some arbitrary perturbations @xmath254 in two different ways , namely , using a finite - difference approximation ( with step size @xmath255 ) and using the inner product of the adjoint - based gradient with the perturbation @xmath254 , namely riesz representation , and then examining the ratio of the two quantities , i.e. , @xmath256 } { \int_{\i } \bnabla^{l_2 } \j_i ( r ) \ , g'_i(r ) \ , dr } , \quad i=1,2 , \label{eq : kappa}\ ] ] for a range of values of @xmath255 . if the gradient @xmath257 is computed correctly , then for intermediate values of @xmath255 , @xmath258 will be close to the unity . remarkably , this behavior can be observed in figures [ fig : kappa]a , b corresponding to problems @xmath155 and @xmath156 over a range of @xmath255 spanning about 8 orders of magnitude . the quantity shown in figures [ fig : kappa]a , b is latexmath:[$\log\| \kappa_i(\epsilon ) - 1 digits to which the two ways to evaluate @xmath253 in agree . furthermore , we also observe that refining the resolution @xmath260 of the time interval @xmath182 $ ] yields values of @xmath258 closer to the unity . the reason is that in the `` optimize - then - discretize '' paradigm adopted here such refinement of the discretization leads to a better approximation of the continuous gradient . as can be expected , the quantities @xmath258 deviate from the unity for very small values of @xmath255 , which is due to the subtractive cancellation ( round - off ) errors in finite - differencing , and also for large values of @xmath255 , which is due to the truncation errors , both of which are well - known effects . in this section we focus on the effect that the choice of initial guess @xmath261 , @xmath81 , has on the reconstructed functions @xmath149 and @xmath150 . we note that , given the nonlinearity of system , optimization problems @xmath155 and @xmath156 may be nonconvex and optimality conditions and characterize minimizers which are only _ local_. thus , different initial guesses may in principle give rise to different reconstructions and convergence to a global minimum can not be a priori assured . we investigate this issue in figures [ fig : gig]a and [ fig : gig]b where we show the reconstructions obtained , respectively , in problems @xmath155 and @xmath156 using different initial guesses generally much worse than mean - field model used in section [ toc : results ] . as regards problem @xmath155 , we note in figure [ fig : gig]a that accurate reconstructions are obtained using even relatively poor initial guesses @xmath162 . on the other hand , in figure [ fig : gig]b we see that in problem @xmath156 the reconstruction fails for a less accurate initial guess @xmath220 . these two examples are representative of the behavior we generally observed in our calculations and we conclude that problem @xmath155 appears more robust with respect to the choice of the initial guess than problem @xmath156 . we also noted that in both problems convergence tends to be more sensitive to the values assumed by the initial guesses @xmath162 and @xmath220 at @xmath124 and @xmath135 than to their behavior for intermediate values of @xmath25 . results from figure [ fig : gig ] are corroborated by the corresponding histories of the cost functionals in figures [ fig : jig]a and [ fig : jig]b . we note that in the case of the poorest initial guess in problem @xmath156 , cost functional @xmath262 reveals hardly any decrease with the iterations at all . while in the cases of successful reconstructions the cost functionals @xmath263 and @xmath262 drop over several orders of magnitude , they never attain values lower than @xmath264 . this is a consequence of the phase - dependent behavior of the measurements which can not be resolved using ansatz in the form , cf . assumption [ ass1](a ) , see also the inset in figure [ fig : ra1a3]a . inverse problems of the type considered here often tend to be ill - posed , in the sense that small perturbations to the data , for example due to noise , may result in significant changes in the computed solution . suitable regularization , for instance , using tikhonov s technique @xcite , may be required to stabilize the solution procedure in such situations . to focus attention in the present study , we concentrated on the structure of the gradients and did not investigate the effect of noise on the reconstructions , hence such regularization was not necessary . we refer the reader to @xcite for a thorough analysis of regularization applied to a related reconstruction problem . in this section we propose some physical interpretation of the numerical reconstruction results from section [ toc : results ] . we note in figures [ fig : g123 ] and [ fig : ra1a3 ] that the identified phase - invariant oscillation model is in fact in remarkably good agreement with the data obtained from the solution of the navier - stokes equation . a small difference between the model and the data is visible as wiggles due to the second harmonic present in the measurements which violates the phase - invariance assumed in our model ansatz , cf . assumption [ ass1](a ) . this difference can be easily removed by a simple pre - processing of the measurement data . referring to appendix [ toc : pod ] , we note that the pod eigenvalue @xmath265 , representing the variance of @xmath22 , is larger than the eigenvalue @xmath266 which represents the variance of @xmath23 . the following rescaling transformation [ eqn : a1a2rescaling ] @xmath267 ensures equipartition of energy in the new variables @xmath268 and @xmath269 while conserving the total energy in both modes . this rescaling effectively removes the second harmonics from @xmath270 , @xmath81 , which could be used as new inputs for the reconstruction . however , we refrained from applying in the computations reported in section [ toc : results ] in order to show the power of the proposed identification method to deal with data which can not be perfectly matched by the model . another observation concerning figure [ fig : g123 ] is the significant deviation of reconstructed functions @xmath149 and @xmath150 from the parabolic mean - field relations . evidently , higher - order corrections , such as @xmath271 , @xmath272 , etc . , are required for a better agreement between the identified propagators @xmath273 , @xmath81 , and the expansions used in the mean - field model . the information which higher - order terms ought to be included in the model as opposed to an a priori fixed polynomial expansion used typically in model identification is therefore the unique advantage of the proposed identification strategy . we note that odd powers of @xmath25 can be excluded by phase - invariance considerations . in addition , with our reconstruction method we were able to determine more accurate values of the instability growth rate at the origin and the oscillation frequency at the limit cycle than used in mean - field model . we stress that in fact such seemingly insignificant modifications of the structure of the reduced - order model may already affect its utility for various control applications . these results also shed light on the validity of mean - field model . initially , the mean - field theory @xcite was derived to be valid near the onset of the oscillation only . we probed the applicability of this model by applying it at a reynolds number @xmath274 which is more than twice the critical value of @xmath61 . hence , the deviation of @xmath149 and @xmath150 from does not invalidate the mean - field theory . one reason for this deviation is the change of the structure of the vortex street during the transient . the optimal oscillatory modes deform from the stability eigenmodes into the pod modes while the fluctuation center moves upstream and the frequency and wavenumber increase @xcite . similarly , the mean - field correction ( the shift mode , cf . ) changes during the transient @xcite which has a noticeable effect on the mean - field model @xcite . finally , the results concerning the identified descriptor system are also relevant to the empirical 9-dimensional galerkin model accounting for the base - flow variation and for the first four harmonics @xcite . the initial exponential growth of the first harmonic is limited by the base - flow variation ( which reduces the production of fluctuation energy ) and , to a lesser extent , by the energy transfer from the first into higher harmonics . the energy transfer may be accounted for by an energy - dependent eddy viscosity in the mean - field system . under certain assumptions ( see , e.g.@xcite ) , a generalized landau equation @xmath275 can be derived , where @xmath95 denotes the growth rate near the fixed point @xmath124 and @xmath109 , @xmath103 characterize the damping from the @xmath276-th and from higher harmonics , respectively . by carefully comparing the 9-dimensional galerkin model with the identified phase - invariant system it may be therefore possible to determine the values of @xmath109 and @xmath103 , or even to correct the powers of the new terms in . a complete derivation and an in - depth discussion of this problem is outside the scope of the present study . we have proposed and validated a novel method for model identification which is an adaptation of an approach already used in the context of systems described by pdes @xcite . as indicated in figure [ fig : f ] , we depart from the traditional approach of ( 1 ) characterizing the propagator of the dynamical system in a parameter space and then ( 2 ) performing a parameter identification . thus , arbitrary polynomial expansions of the propagator may be performed a posteriori ( following the solution of the optimal reconstruction problem ) at a practically vanishing cost . in addition , the performance of parametric models may easily be assessed and , if necessary , improved by introducing additional terms motivated by the form of the reconstructed constitutive relation . the method is applied to a three - dimensional descriptor system with three a priori undetermined relations describing the fluctuation growth , frequency and mean - field correction as functions of the fluctuation energy . as a benchmark problem we chose the onset of laminar von krmn vortex shedding behind a circular cylinder . results of a direct numerical simulation are transcribed into the mode amplitudes of a minimal 3-state galerkin model @xcite which are then captured with remarkable accuracy by our identified descriptor system . the form of the reconstructed system is marked by a noticeable departure from the mean - field models and may therefore guide the refinement of the latter by inclusion of higher - order terms . we emphasize that the usefulness of reduced - order models for flow control applications may in fact depend on such differences . as regards future research directions , while the present results offer a proof of the concept for the proposed approach based on a rather well - understood example , the key question is extension of this method to the identification of models with more complicated structure featuring , for example , multiple time scales , state space of a higher dimension , coexistence of several oscillation frequencies , non - trivial phase dependence and higher - dimensional inertial manifolds . as regards the first issue , one can consider a modification of our model problem with assumption [ ass1](b ) revised to allow @xmath277 to be a `` fast '' variable . while in such setting our computational approach would formally remain unchanged ( except that problem @xmath204 would be replaced with a problem similar to @xmath155 or @xmath156 ) , it is interesting how it would actually perform in practice . dealing with some of the other aspects will require formulation of the reconstruction problems in terms of propagator functions depending on more than just one state variable ( @xmath25 in the examples considered in the present study ) . this will , in turn , lead to a number of interesting questions at the level of numerical analysis and scientific computing related to the evaluation of the cost functional gradients . an emerging application which involves some of the aforementioned extensions is related to the question of optimal parametrization of subgrid turbulence representations which is an important open problem in theoretical fluid mechanics @xcite . in the context of galerkin reduced - order models , it may take the form of an additional dissipative term with the magnitude proportional to an `` eddy viscosity '' @xmath278 @xmath279 where @xmath280 represents the stabilizing viscous term of the navier - stokes equation in the galerkin system . in general , this term can be proven to be energy dissipative for all orthonormal systems of modes and a large class of boundary conditions . for some analytical modes , e.g. , the stokes modes , it can be shown that matrix @xmath281 is diagonal and negative - definite . there is abundant evidence @xcite that nonlinear closure strategies perform better as regards stabilization of system . assuming @xmath282 gives rise to an identification problem analogous to @xmath155 and @xmath156 , and one can use the algorithm described in section [ toc : computation ] to determine optimal closure strategies leading to the best possible reconstruction of the available data . preliminary identification results already obtained based on a reduced - order model with dimension @xmath283 applied to a complex mixing - layer flow are quite encouraging and reveal some nontrivial physical insights . they will be reported in the near future upon completion of the study . we also remark that a surprisingly large set of modelling and control problems can be cast in a similar form of function identification of a descriptor system @xmath284 for reasons of simplicity , let us assume that function @xmath11 is known and that function @xmath285 needs to be determined . if @xmath286 characterizes the slow modes , then @xmath287 represents the inertial manifold to be identified from a given system trajectory @xcite . if @xmath286 represents high - frequency components , such as the parameters of a subgrid turbulence representation described above , then their functional dependence on the state variable @xmath288 may also be inferred with our approach . on the other hand , if @xmath286 denotes the actuation amplitudes , as in numerous wake flow stabilization studies @xcite , then @xmath289 represents a full - state feedback control law . in principle , this control law may as well be identified from desired trajectories @xmath14 . the authors acknowledge the funding and excellent working conditions of the chair of excellence closed - loop control of turbulent shear flows using reduced - order models ( tucorom ) of the french agence nationale de la recherche ( anr ) and hosted by institute pprime . the first author is , in particular , grateful for the hospitality of this chair of excellence at institute pprime where most of this work was carried out . we also thank the ambrosys ltd . society for complex systems management and the bernd noack cybernetics foundation for additional support . we appreciate valuable stimulating discussions with markus abel , robert niven , michael schlegel and gilead tadmor as well as the local tucorom team : jean - paul bonnet , laurent cordier , thomas duriez , peter jordan , vladimir parezanovic and andreas spohn . in this appendix we describe the proper orthogonal decomposition ( pod ) employed in section [ toc : galerkinexpansion ] to construct a low dimensional model from the simulation data . it is closely related to other techniques of data analysis known as the principal component analysis or , in the discrete setting , the singular value decomposition . the starting point are snapshots of the velocity field @xmath290 , @xmath291 , @xmath292 , cf . . these snapshots are sampled at times @xmath293 uniformly spaced over one period of oscillation and form a statistically representative ensemble for the considered first and second moments . the goal is to construct a least order galerkin expansion @xmath294 with base mode @xmath295 and @xmath296 space - dependent expansion modes ( basis functions ) @xmath297 with the corresponding mode amplitudes @xmath298 which will result in a minimum - norm average residual @xmath299 of the snapshot ensemble ( see , e.g. , @xcite ) . the base mode is necessary so that galerkin expansion satisfies inhomogeneous boundary conditions for arbitrary mode amplitudes . for instance , expansion captures the prescribed oncoming flow velocity regardless of the values of @xmath300 . the galerkin expansion and its residual are embedded in the hilbert space @xmath301 of square - integrable vector fields . the inner product of two elements @xmath302 is defined by @xmath303 where @xmath304 denotes the standard euclidean inner product and @xmath305 an infinitesimal volume element of the domain @xmath71 . the associated norm thus is @xmath306 we search for empirical modes @xmath80 , @xmath307 , where @xmath308 , which minimize the average residual of the galerkin expansion of the snapshots @xmath309 with optimal mode amplitudes @xmath310 in the sense of the @xmath198 norm , i.e. , @xmath311 this problem is solved by the snapshot pod @xcite and the base mode is the mean of the snapshots @xmath312 the pod modes arise from the correlation matrix @xmath313 of the snapshot fluctuations @xmath314 we note that @xmath315 is a positive semi - definite grammian matrix . hence , the eigenvalue problem @xmath316 yields an orthonormal set of real eigenvectors @xmath317^t $ ] , @xmath318 , with non - negative eigenvalues which can be ordered as @xmath319 the last equality arises from the fact that @xmath320 vectors span a subspace of maximum dimension @xmath321 . hence , @xmath320 snapshots define only @xmath321 pod modes and the corresponding amplitudes . these are given by @xmath322 the pod modes form an orthonormal basis in @xmath68 , @xmath323 , @xmath324 , while their amplitudes have vanishing means @xmath325 , @xmath326 , and diagonal second moments @xmath327 , @xmath324 . the average is to be understood in terms of the snapshot ensemble ( see , e.g. , ) . the eigenvalue @xmath328 can be interpreted as twice the fluctuation energy contained the @xmath329-th mode . by the one - dimensional embedding result @xmath330 @xcite , we note that the reconstructed function will be hlder - continuous with @xmath331 . thus , it will not meet the assumptions of the picard - lindelf theorem @xcite , and in principle will ensure existence only , without uniqueness , of solutions of equation . in order to ensure the lipschitz - continuity ( @xmath332 ) of @xmath44 , which would also guarantee uniqueness of solutions of , we would need to reconstruct @xmath44 as an element of sobolev space @xmath333 , because then @xmath334 . while there are no fundamental difficulties here ( we would need to replace inner product with the corresponding definition in @xmath333 ) , we will refrain from this in the actual computations in section [ toc : results ] in order to keep the approach as simple as possible . nevertheless , in all problems we treated with the proposed approach the reconstructed functions possessed the lipschitz regularity which was verified a posteriori by performing suitable grid - refinement studies . in this appendix we present a derivation of perturbation equation . obtaining this equation is made somewhat more involved by the fact that the perturbation variable @xmath44 is itself a function of the state magnitude , i.e. , @xmath335 . we assume here that only @xmath44 is perturbed while @xmath45 remains fixed , with the opposite case leading to essentially the same calculations . by substituting , respectively , @xmath336 and @xmath337 into equation , we obtain @xmath338 and @xmath339 , where @xmath340 and @xmath341 are the corresponding solutions and @xmath342 , @xmath343 . taking the difference of these two equations and defining @xmath344 . we obtain @xmath345 where we also set @xmath346 in the first term on the rhs . as regards the terms denoted @xmath347 in , they are transformed as follows using the fundamental theorem of calculus for line integrals and the change of variables @xmath348 for @xmath349 $ ] @xmath350 where we also denoted @xmath351 . the integrand expression on the rhs in is then expanded in the taylor series around @xmath352 @xmath353 where the component notation was used for clarity with @xmath354 , @xmath355 and @xmath356 denoting the components of vectors @xmath357 , @xmath358 and @xmath176 . plugging expansion into expression we obtain @xmath359 \ , \bxi ' + \o\left ( \|\bxi'\|^2 \right ) . \end{aligned } \label{eq : a2}\ ] ] noting that term @xmath360 in transforms in an analogous way to @xmath347 in , using these results in , assuming smallness of @xmath176 and dropping terms of order quadratic and higher we finally arrive at perturbation equation . j. watson . on the non - linear mechanics of wave disturbances in stable and unstable parallel flows . the development of a solution for plane poiseuille flow and for plane couette flow . , 9:371389 , 1960 . m. morzyski , w. stankiewicz , b. r. noack , f. thiele , and g. tadmor . generalized mean - field model for flow control using continuous mode interpolation . in _ 3rd aiaa flow control conference _ , san francisco , ca , usa , 5 - 8 june 2006 , 2006 . invited aiaa - paper 2006 - 3488 . j. gerhard , m. pastoor , r. king , b. r. noack , a. dillmann , m. morzyski , and g. tadmor . model - based control of vortex shedding using low - dimensional galerkin models . in _ 33rd aiaa fluids conference and exhibit _ , orlando , florida , usa , june 2326 , 2003 , 2003 . paper 2003 - 4262 .	we propose a general framework for parameter - free identification of a class of dynamical systems . here , the propagator is approximated in terms of an arbitrary function of the state , in contrast to a polynomial or galerkin expansion used in traditional approaches . the proposed formulation relies on variational data assimilation using measurement data combined with assumptions on the smoothness of the propagator . this approach is illustrated using a generalized dynamic model describing oscillatory transients from an unstable fixed point to a stable limit cycle and arising in nonlinear stability analysis as an example . this 3-state model comprises an evolution equation for the dominant oscillation and an algebraic manifold for the low- and high - frequency components in an autonomous descriptor system . the proposed optimal model identification technique employs mode amplitudes of the transient vortex shedding in a cylinder wake flow as example measurements . the reconstruction obtained with our technique features distinct and systematic improvements over the well - known mean - field ( landau ) model of the hopf bifurcation . the computational aspect of the identification method is thoroughly validated showing that good reconstructions can also be obtained in the absence of of accurate initial approximations . * keywords : * hydrodynamic instabilities , reduced - order models , mean - field models , variational data assimilation , adjoint - based optimization , * ams subject classifications : * 93a30 , 65k10 , 76d25
the _ einstein _ satellite discovered x - ray emitting gas halos around luminous early type galaxies ( eg . forman , jones , & tucker 1985 ) , forcing the rejection of previous assumptions that elliptical galaxies are gas poor systems . the x - ray surface brightness distribution was found to closely follow the optical image ( trinchieri , fabbiano , & canizares 1986 ) ; however the x - ray luminosity , and hence the mass of the hot gas , showed large scatter from galaxy to galaxy ( eg . canizares , fabbiano , & trinchieri 1987 ) . the estimated cooling time of the hot gas is less than the hubble time ( trinchieri et al . 1986 ) , however cooling may be balanced by heating mechanisms such as supernova explosions . arnaud et al . ( 1992 ) found that the iron mass in the intracluster medium ( icm ) is directly proportional to the optical luminosity from early type galaxies in clusters of galaxies . this leads one to suspect that metals in the icm originate in early type galaxies and that there might be a link , in the form of early star formation and galactic winds , between the icm and metal abundances in early type galaxies . due to the poor energy resolution of previous satellites , however , critical parameters for investigating the nature of the hot gas in early type galaxies , such as the gas temperature , mass , and metal abundance have not been well - constrained . accordingly , several important issues including the origin , heating , and fate of the hot gas , and the effect of early type galaxies on the icm , are not well understood . _ asca _ ( tanaka et al . 1994 ) is the first satellite with sufficient energy resolution and large effective area over the broad 0.4 to 10 kev bandpass to determine these physical parameters . with this satellite , awaki et al . ( 1994 ) analyzed 3 early type galaxies and found that the temperature and metal abundance of the hot gas are about 0.8 kev and 0.4 solar , respectively . the inferred metal abundance is lower than expected from theoretical models , and lower than those determined at other wavelengths . accordingly , the _ asca _ observations raise a serious challenge to standard models for the evolution of galaxies and clusters of galaxies . this mystery deepened with the discovery of even lower metallicities ( less than 0.2 solar ) in two other elliptical galaxies ( loewenstein et al . 1994 ) , and in the outer regions of ngc4636 ( mushotzky et al . 1994 ) . matsushita et al . ( 1994 ) discovered , analyzing 5 early type galaxies including the sample of 3 in awaki et al . ( 1994 ) , a hard x - ray component at energies above a few kev in addition to the hot gas component dominating below about 2 kev . in order to perform plasma diagnostics of the hot gas component , it is essentially important to estimate and remove the hard component from the x - ray spectra of early type galaxies . if the hard component is due to the integrated emission of stellar sources , as suggested by matsushita et al . ( 1994 ) , its luminosity should be proportional to optical luminosity . since the samples in previous studies are too limited to fully investigate all the issues mentioned above , we have systematically studied twelve early type galaxies , including the samples of awaki et al . ( 1994 ) and matsushita et al . ( 1994 ) as well as newly observed galaxies . these comprise the largest sample of _ asca _ spectra of early type galaxies yet investigated . the galaxies we have selected for this study are ngc4472 , ngc4406 , ngc4636 , ngc4649 , ngc499 , ngc507 , ngc720 , ngc4374 , ic4296 , ngc4382 , ic1459 and ngc4365 , which are listed in table 1 with their optical parameters . the sample , which includes 7 performance and verification ( pv ) phase targets and 3 first - round guest investigator observations by the first author ( hm ) , was principally selected on the basis of soft x - ray fluxes measured with the _ einstein _ satellite in order to obtain useful spectra in reasonable exposure times . in addition to this criterion , a number of low x - ray luminosity galaxies were intentionally selected in order to detect and study possible differences between gas - rich and gas - poor galaxies . furthermore , we selected our sample to include both galaxies in dense and tenuous gaseous environments in order to be able to study possible interaction between the icm and interstellar medium ( ism ) . all the galaxies were observed with two solid - state imaging spectrometers ( sis0 and sis1 ) and two gas imaging spectrometers ( gis2 and gis3 ) at the foci of 4 thin - foil x - ray mirrors ( xrt ) on board the _ asca _ satellite . details of the instruments can be found in burke et al . ( 1991 ) , ohashi et al . ( 1996a ) , makishima et al . ( 1996 ) , and serlemitsos et al . ( 1995 ) , while a general description of _ asca _ can be found in tanaka et al . ( 1994 ) . in table 2 , we summarize the _ asca _ observation log of our sample galaxies . the data were screened with the standard selection criteria : data affected by the south atlantic anomaly , earth occultation , and regions of low geomagnetic rigidity were excluded . we also eliminated contamination by the bright earth , removed hot and flickering pixels for the sis data , and applied rise - time rejection to exclude particle events for the gis data . after the above data screening , we extracted x - ray spectra from circular regions centered on each galaxy with a uniform physical extraction radius of 38 kpc the largest radius completely contained within the sis field - of - view for all galaxies . the background spectra were obtained from source free regions around each galaxy . in order to utilize the @xmath1 technique , we rebinned the x - ray spectra to contain at least 20 counts in each spectral bin , then simultaneously fitted the spectra from the sis0 , sis1 , gis2 and gis3 , with a common normalization factor but separate response functions for each detector , using the xspec ( version 8.50 ) spectral fitting package . awaki et al . ( 1994 ) reported that the sis spectra of bright early type galaxies , ngc4636 , ngc4472 and ngc4406 exhibit emission lines from highly ionized fe , si , and s atoms . we further confirmed the presence of these emission lines in the spectra of the additional bright galaxies ngc4649 , ngc499 , ngc507 and ngc4374 . therefore we inferred that thin hot plasma emission is prevalent in bright early type galaxies and , hence , attempted to fit the x - ray spectra with a single temperature thermal plasma model ( raymond & smith 1977 ; hereafter rs model ) modified by interstellar absorption . the free parameters were the plasma temperature , metal abundance , photoelectric absorption column density , and normalization . the abundances are collectively varied keeping the relative ratio at the solar value ( anders & grevesse 1989 ) . this model , however , failed to reproduce the observed spectra except that of ngc499 . from all galaxies , systematic excess emission is found above the 4 kev band . a typical example , ngc4649 , is shown in figure 1 . thus we found the presence of hard x - ray emission , as previously reported in several individual galaxies ( kim , fabbiano & trinchieri 1992 ; matsushita et al . 1994 ) , to be a ubiquitous property of elliptical galaxies . in figure 2 we compare the x - ray spectra of the two elliptical galaxies ngc4636 and ngc4365 ; although they have nearly identical optical blue luminosities , the former is x - ray bright and the latter is x - ray faint . the difference in x - ray flux is confined to the soft x - ray band : the x - ray spectrum of ngc4636 below 3 kev is nearly two orders of magnitude brighter than that of ngc4365 . on the other hand , the x - ray spectra above the 4kev band are indistinguishable , as might be expected based on the identical optical blue band magnitudes . thus we compared the x - ray spectra above the 4 kev band for the entire sample , and found no evidence of significant differences in spectral shape , although the absolute fluxes vary from galaxy to galaxy . accordingly , we assumed a thermal bremsstrahlung model for the hard component to estimate the temperature variation using the data above 4 kev . since the statistics are limited , however , we could only determine a lower limit on the temperature of the bremsstrahlung model of about 2 kev . since the gis has higher sensitivity than the sis in the relevant high energy band , we co - added all of the gis spectra in our sample to improve the statistical accuracy . figure 3 shows the composite gis 2 and 3 spectra above the 4 kev band . we fitted a thermal bremsstrahlung model to this spectrum , and obtained an acceptable fit ( @xmath2 of @xmath3 ) with the best - fit temperature @xmath4 kev ( hereafter , errors are 90 % confidence unless otherwise stated ) . this composite spectrum is equally well fitted ( @xmath2 of 22.83/31 ) by a power law model with photon index 1.8@xmath5 . therefore , we can not distinguish between the two possible origins thermal and non - thermal of the hard component from the spectrum alone . however the relevant physical parameters , such as the x - ray luminosity ( @xmath6 ) of the hard component , or iron k - shell line flux do not significantly differ whether we use thermal bremsstrahlung or power - law models . thus we assume the thermal - bremsstrahlung model for further discussion . in order to estimate the strength of fe k line emission , we added a gaussian line to the above model . the best - fit temperature is slightly altered to @xmath7 kev , with line center energy and equivalent width of the gaussian of @xmath8 kev and @xmath9 ev , respectively . confidence contours of the line energy and equivalent width at @xmath10 2.71 , 4.61 and 9.21 are plotted in figure 4 . adding the line decreased @xmath1 from 22.36 to 18.50 , which is not statistically significant ( malina , lampton & bowyer 1976 ) . therefore we can conclude that the composite hard component spectrum requires no significant emission lines . since the presence of both emission lines from highly ionized heavy elements in the low energy band and a hard x - ray tail at the high energy band seem to be common in the galaxies in our sample , we fitted two - component models to all broad band spectra . the models consist of a thermal bremsstrahlung component with a fixed temperature of 12 kev determined from the composite hard band spectrum , plus a thin thermal model ( rs model ) for the soft band emission . the column density of the hard component was constrained to be the same as that of the rs model . the free parameters were the temperature and metal abundance of the rs model , the column density , and normalizations for both the rs and bremsstrahlung models . the best - fit parameters and 90% confidence errors are listed in table 3 . we generally found reasonable fits with this unified model , however two galaxies ngc4636 and ngc507 exhibit unacceptably large values of @xmath1 . although these two galaxies may require more complicated modeling ( ngc4636 was found to have both an abundance gradient and an extra centrally concentrated cool component by mushotzky et al . 1994 ) , more detailed study of individual galaxies is beyond the scope of this paper . the best - fit rs abundances and temperatures are in excellent agreement with those derived from _ rosat _ pspc observations ( davis & white 1996 ; kim & fabbiano 1995 ) , except for ngc4472 where the _ asca _ parameters are consistent only with the central pspc parameters ( forman et al . significant excess absorption is required by the _ asca _ , but not the pspc , spectra of ngc4472 , ngc4406 , and ngc4649 . this discrepancy is likely due either to remaining inaccuracies in the _ asca _ low energy response , or to the inadequacy of the simple foreground screen model that we employ for the absorption . many authors have pointed out systematic differences in the best - fit parameters between models using different plasma emission codes . we thus examined this possible systematic and model - dependent uncertainty by comparing the rs and meka models ( mewe , r. , gronenschild , e.h.b.m . , and van den oord , g.h.j . 1985 ) . in figures 5a d , we show the correlation plots of best - fit parameters between the rs and meka models : the column density ( a ) , temperature ( b ) , metal abundance ( c ) , and x - ray luminosity ( d ) . the best - fit column densities are found to be consistent within the errors , while best - fit temperatures in the meka model are almost 20 % lower than the rs model . the metal abundances in the meka model generally have larger values than the rs model , although the scatter in the correlation plot is large . as for the luminosity , both models give essentially the same best - fit value . thus we confirmed the existence of plasma model - dependent systematic and non - systematic errors . however these do not largely effect the following discussion , hence we refer to the best - fit values using the rs model . figure 6 shows the relation between the x - ray luminosity ( @xmath6 ) of the hard component in the 0.54.5 kev band and the blue band luminosity ( @xmath11 ) . the uncertainties in @xmath6 of the hard component are typically @xmath1230% . the solid line is the best - fit relation between the blue luminosity and the 0.54.5 kev band luminosity found for bulge - dominated spiral galaxies ( canizares et al . we found that @xmath6 of the hard component is proportional to @xmath11 , and that the constant of proportionality is nearly the same as found in spiral galaxies : @xmath13 for the hard component in early type galaxies is generally consistent with that of bulge - dominated spiral galaxies like m31 . the excess hard flux in the radio galaxies ic1459 and ic4296 probably can be explained by x - ray emission from agn ; indeed our analysis of archival pspc data from ic1459 indicates the presence of an unresolved hard component . furthermore , we found that the composite _ asca _ gis spectrum above the 4kev band ( section 3 ) resembles the x - ray spectra of bulge - dominated spiral galaxies which can be fitted with a thermal bremsstrahlung model with @xmath14 kev ( eg . makishima et al . 1989 ) . these facts suggest that early type galaxies contain a hard x - ray component attributable to the same origin as x - rays in bulge - dominated spiral galaxies . since the x - ray emission from bulge - dominated spirals is thought to arise from a population of low - mass x - ray binaries ( lmxb ) , we suspect that the primary hard component in early type galaxies is also due to the integrated emission of an ensemble of lmxb . matsushita et al . ( 1994 ) came to the same conclusion based on a more limited ( 5 galaxies ) sample ; they also discovered that the hard component is generally extended which argues against an agn power - law origin for the hard component . with a larger galaxy sample than matsushita et al . ( 1994 ) , we therefore confirm that both the shape and relative magnitude of the hard component in early type galaxies is consistent with that in bulge - dominated spiral galaxies , strengthening the suggestion that the hard x - ray component in early type galaxies is due to lmxb . although lmxb usually exhibit a weak 6.7 kev iron line , we found no significant iron line in the early type galaxies . the 90 % confidence upper limit on the equivalent width of the fe k@xmath15 line of 500 ev is too large to place any strong constraints on the lmxb scenario . emission lines from highly ionized fe , si , and s are detected from the 7 galaxies in our sample with the best statistics ; non - detection of emission lines for the other galaxies is consistent with their having a lower signal - to - noise ratio . the existence of these emission lines is direct evidence for the prevalence of thin hot interstellar plasma in bright early type galaxies . we now discuss the nature of this plasma using our results , as well as referring to published optical data . because @xmath6 of the hard component is generally proportional to @xmath11 , the large dispersion in the total @xmath6-@xmath11 diagram ( canizares et al . 1987 ) is mainly due to a spread in @xmath6 of the soft component . this is illustrated in figure 7 where @xmath6 of the soft component is plotted versus @xmath11 . since , ignoring the mostly small variations in temperature , @xmath6 of the soft component is proportional to the volume emission measure ( @xmath17 ) , and @xmath18 , where @xmath19 and @xmath20 are the densities of electrons and protons , respectively , and @xmath21 and @xmath22 are , respectively , the volume and mass of the hot gas , the large dispersion in @xmath6 found in figure 7 indicates the presence of a wide mass range of hot gas in early type galaxies . figure 8 is a plot of the volume emission measure ( @xmath17 ) versus the temperature of the hot gas . for comparison , we show the data for clusters of galaxies in the same figure with plus symbols ( hatsukade 1989 ) . the spearman rank correlation coefficient between @xmath17 and temperature for early type galaxies is estimated to be 0.43 ; therefore , we can not conclude that there is a strong correlation between temperature and mass of the hot gas . the line in the figure shows the best - fit relation for the cluster data . what is clear is the fact that most of our galaxy sample is located well below the cluster correlation line , and that @xmath17 drops steeply with @xmath16 from rich cluster all the way down to galaxy scales . since for a constant gas mass as a fraction of gravitational mass @xmath17 @xmath23 , it is clear that early type galaxies are far more gas poor than clusters . figure 9 presents the relation between the stellar velocity dispersion and the hot gas temperature . the velocity dispersions of the galaxies in clusters are shown with plus signs ( hatsukade 1989 ) . the solid line represents @xmath24 , where @xmath25 , and @xmath26 , @xmath27 , @xmath28 , @xmath29 and @xmath30 are mean molecular weight ( we assumed it to be 0.6 ) , proton mass , stellar ( or galaxy ) velocity dispersion , boltzmann constant , and temperature , respectively . @xmath24 indicates that the thermal energy of the hot gas per unit mass is equal to the stellar ( or galactic ) kinetic energy per unit mass . the spearman rank correlation coefficient for early type galaxies between @xmath28 and temperature is estimated to be 0.54 , which is not significant . thus we can not strongly conclude that there is a positive correlation between these two parameters which would indicate a correlation between the depth of the gravitational potential and the temperature of the hot gas as expected for equilibrium systems . however , all our sample galaxies show higher temperatures than would be expected based on the best - fit cluster sample correlation line ; _ i.e. _ @xmath0 is less than 1 for all of the early type galaxies in our sample compared to approximately 1 for clusters of galaxies . in other words , the thermal energy of the hot gas per unit mass is larger than the stellar kinetic energy per unit mass . figure 10 shows the relation between the velocity dispersion and @xmath0 . the mean value of @xmath0 , weighted using relative errors , is about 0.5 . if the stars and gas in early type galaxies are both isothermal and in hydrostatic equilibrium , @xmath31 , where @xmath32 and @xmath33 are the densities of hot gas and stars in the galaxy , respectively ( eg . sarazin 1986 ) . trinchieri et al . ( 1986 ) have found that for bright early type galaxies , the x - ray and optical surface brightness tend to follow each other . since the x - ray brightness profile traces @xmath34 and the optical profile traces @xmath33 , this implies that @xmath0 is about 0.5 ( fabbiano 1989 ) , consistent with our direct measurement of @xmath0 . @xmath35 is expected if cooling ( proportional to @xmath34 ) is approximately balanced by heating by stellar winds , gravitational compression in an inflow , or type ia supernovae that all scale as the local stellar density . assuming that the present - day stellar mass loss rate is about @xmath36 and has decreased with time as @xmath37 ( ciotti et al . 1991 ) , we see that the typical mass of hot gas observed in early type galaxies ( @xmath38 ) is much less than the total accumulated over the hubble time ( @xmath39 ) . this fact in conjunction with @xmath35 and the short cooling times of the hot gas in the central regions ( @xmath40 10@xmath41 yr ; fabbiano 1989 ) implies that either ( 1 ) there is some heating mechanism that balances cooling and has driven much of the stellar mass loss out of the galaxy , or ( 2 ) an inflow is occurring where the primary heating mechanism is gravitational compression and where there must be some sink for gas cooling below x - ray emitting temperatures . one may argue that a plausible heating mechanism is type ia supernova explosions . we consider a simple model where the energy of supernova explosions is constant ( @xmath42 erg ) and the energy deposited per unit gas mass uniform over the entire galaxy . let @xmath43 be the `` excess temperature '' the difference between the temperature of the hot gas and that corresponding to the stellar velocity dispersion , @xmath44 since we are considering emission from a standard volume , the mass of the hot gas is proportional to the square root of @xmath17 ( section 4.1.1 ) , and the supernova rate thought to be proportional to @xmath11 . if heating by type ia supernovae is important , we would expect a negative correlation between @xmath43 and @xmath45 because the rise in gas temperature should be smaller as the gas mass per supernova becomes larger . figure 11 shows the plot of @xmath45 vs. @xmath43 . since the spearman rank correlation coefficient is 0.40 , we can conclude that there is no negative correlation . this contradicts simple models where the heating source is proportional to @xmath11 and the efficiency of heating is uniformly high , but rather suggests that the efficiency of energy transfer to the hot gas may increase as the gas mass per star becomes larger . a negative correlation between @xmath45 and fe abundance would also be predicted by a simple snia - dominated heating model ( see below ) , and is likewise not seen . the importance of heating by type ia supernovae can generally be ruled out on the basis of the observed constraints on the iron abundance , since each explosion injects @xmath46 of fe into the hot gas . this means that increasing the gas temperature by @xmath47 kev will enhance the fe abundance by @xmath48 in solar units . even if we attribute all of the fe in the hot gas to supernova explosions an extreme assumption considering the expected contribution from stellar mass loss ( section 4.2.3 ) supernova heating can not explain the observed values of @xmath43 in most of our sample . figure 12 shows the relation between the abundance and the temperature . we can see no clear correlation between these parameters , although the lowest metallicities are found in the galaxies with the coolest ism . if we use the meka model , these conclusions are not essentially altered . for the galaxies whose metal abundances are well determined , the abundances , without exception , are found to be less than half solar . these values are considerably lower than theoretical expectations , and those determined at optical wavelengths . an abundance gradient ( mushotzky et al . 1994 ) may partially account for the discrepancy between optical abundances , which are determined in the central region of the galaxy , and x - ray abundances measured over a region extending to many effective radii . significantly sub - solar metallicities have also been observed with _ asca _ in a variety of sources , including stellar coronae ( eg . singh , drake , & white 1995 ) , starburst galaxies . awaki et al . 1995 ; tsuru et al . 1996 ) , and supernova remnants ( snrs ) ( eg . hayashi et al . 1996 ) . we think it is worth noting that some snrs in the large magellanic cloud sufficiently evolved to sweep up large amounts of ism exhibit x - ray spectra which closely resemble those of early type galaxies ( hayashi et al . 1996 ) . improvements in our understanding of the processes of enrichment by supernovae and stellar mass loss may help solve the puzzle of low abundances if , for example , the gas ejected from supernovae does not mix well into the ism and the gas which is overly dense and metal - rich cools and drops out of the x - ray emitting hot gas ( fujita , fukumoto , & okoshi 1996 ) . arimoto et al . ( 1996 ) suggest that low abundances may be an artifact , since the metal abundance is constrained by fe l emission lines around 1 kev for which the atomic physics parameters are poorly known . however , hwang et al . ( 1996 ) have found that , for hot gas temperatures between 2 and 4 kev , the abundance determined from fe l and fe k lines is fairly consistent ( with abundances determined from the fe l region actually slightly higher ) , and any errors in the abundances due to atomic physics uncertainties that exist in the current codes are unlikely to exceed 30 50 % . therefore plasma model uncertainties may not explain the low abundances , although the increasing complexity of fe l emission for temperatures below 1 kev imply that more detailed analysis than lies within the scope of this paper is needed to solve this problem . we note that the consistency of fe abundances derived independently over a wide range of ionization states in stellar coronae ( eg . drake 1996 ) and the comparably low abundances inferred for ions with more straightforward atomic parameters ( eg . si ; ohashi et al . 1996b ) implies that it is unlikely that the abundances derived here are underestimated by large factors . weak constraints on oxygen abundances are available only for the 3-brightest early type galaxies ( ngc4472 , ngc4406 , and ngc4636 ) ; other galaxies have insufficient signal - to - noise ratios . the best fit oxygen abundances for the 3 galaxies we derive are consistent with those found by awaki et al . ( 1994 ) , roughly equal to the iron abundances . the oxygen abundances are not determined to sufficient accuracy to place significant constraints on the relative contributions of snii and snia to the heavy metal production in the observed galaxies . our sample includes galaxies in both relatively dense and tenuous intergalactic environments . ngc4636 , for example , is very far from the center of the virgo cluster and no nearby icm x - ray emission is seen , while ngc4472 , the central galaxy of a subgroup in the virgo cluster , is surrounded by significant icm x - ray emission . ngc720 , at the center of the poor group lgg38 ( garcia 1993 ) , on the other hand , does not show effects from either ram pressure or tidal distortion , hence is regarded as rather isolated ( buote & canizares 1996 ) . for all the correlations shown in figures 6 12 , we see no systematic differences between the galaxies in dense and tenuous gaseous environments . this indicates that the interaction between the icm and ism is not strong enough to manifest itself in our sample . more comprehensive investigations of such interactions require detailed study of spatial variations in x - ray spectral properties as in iwasawa et al . ( 1996 ) who have found evidence for ram pressure stripping of the ism in ngc4406 , as well as a larger sample spanning a wider range of environments . we have presented results from the uniform analysis of the _ asca _ spectra of 12 early type galaxies . this is the largest such sample to date and , as summarized below , we have uncovered new facts as well as confirmed previous inferences based on much smaller samples . 1 . we found that the x - ray spectra from the 12 galaxies in our sample generally consist of at least two components : a soft component from x - ray emission of hot gas , and a hard component that can be characterized by either a thermal bremsstrahlung model with @xmath49 kev or a power - law model of photon index 1.8 . the consistency of the strength and spectral shape of the hard component with those in bulge - dominated spirals corroborates previous suggestions that it primarily originates as the integrated emission from x - ray binaries . the hard component spectrum requires no significant fe k line emission , and the upper limit on the equivalent width of fe k line emission is 460 ev . we confirmed that the metal abundance of the hot gas is generally less than 0.5 solar . there is no clear correlation between the abundance and temperature of the hot gas , however the lowest abundances are found in the coolest galaxies . there is a large dispersion in the ratio of gas mass ( _ i.e. _ the x - ray luminosity of the soft component ) to optical luminosity , implying a wide variety of gas - dynamical histories . 4 . theories of large - scale structure imply that there is a continuum of dark matter halos spanning galaxy and cluster scales . however , we have found that early type galaxies fall well below the constant gas mass fraction extension of the hottest clusters in the @xmath50 plane ( vem @xmath23 ) , and that the ratio of thermal to kinetic energy per unit mass is , on average , twice that found for clusters of galaxies . this suggests that larger systems are generally more efficient at retaining their hot gas and less efficient at converting gas into stars . the interstellar media of elliptical galaxies are not simply scaled down version of intracluster media , but have been more profoundly effected by feedback from star formation and other gas - dynamical processes . the low fe abundances , as well as the absence of a negative correlation between the `` excess temperature '' , defined as the difference between the hot gas temperature and the temperature corresponding to the stellar velocity dispersion , and the ratio of the square root of the volume emission measure of the hot gas to the blue band luminosity ( an indicator of the mass ratio of hot gas to stars ) argue against type ia supernovae being an important heat source for the ism . we found no systematic differences in the physical parameters of the hot gas between galaxies in dense and tenuous intracluster environments , although our ability to test this is limited . this indicates that interaction between the intracluster medium and interstellar matter may not be strong . in short , _ asca _ analysis of early type galaxies has yielded both expected ( a hard component with the same normalization as in spiral bulges ) and unexpected ( low interstellar abundances ) results . the interrelationships and dispersion among the properties of the hot gas component , and their comparison with those in intracluster media demonstrates the relative importance of astration , mass loss , and gas - dynamics in determining the evolution of early type galaxies and their ism . any complete theory of galaxy formation must explain the observed trends and diversity of bound distributions of hot gas on galaxy and cluster scales . awaki , h. , tsuru , t. , koyama , k. , & _ asca _ team 1996 , in uv and x - ray spectroscopy of astrophysical and laboratory plasmas , ed . yamashita , k. , and watanabe , t. ( tokyo , japan : universal academy press , inc . ) , 327 lcccccccccc ngc4472(m49 ) & 12 29 46.5 & 07 59 58 & 25.8 & 9.26 & -22.80 & 11.28 & e2 & 315 & 1.6e20 & virgongc4406(m86 ) & 12 26 11.8 & 12 56 49 & 25.8 & 9.71 & -22.35 & 11.10 & e3 & 256 & 2.7e20 & virgongc4636 & 12 42 49.8 & 02 41 17 & 25.8 & 10.37 & -21.69 & 10.84 & e0 - 1 & 217 & 1.9e20 & virgongc4649(m60 ) & 12 43 40.3 & 11 32 58 & 25.8 & 9.70 & -22.36 & 11.11 & e2 & 344 & 2.4e20 & virgongc499 & 01 23 11.6 & 33 27 36 & 94.5 & 12.79 & -22.09 & 11.00 & s0@xmath51 & 235 & 5.3e20 & piscesngc507 & 01 23 40.1 & 33 15 22 & 94.5 & 12.19 & -22.69 & 11.24 & sa(r)0@xmath52 & 306 & 5.3e20 & piscesngc720 & 01 53 00.4 & -13 44 21 & 39.6 & 11.13 & -21.86 & 10.91 & e5 & 224 & 1.4e20 & lgg38ngc4374(m84 ) & 12 25 03.7 & 12 53 15 & 25.8 & 9.91 & -22.15 & 11.02 & e1 & 296 & 2.6e20 & virgoic4296 & 13 36 38.9 & -33 57 59 & 72.8 & 11.42 & -22.89 & 11.32 & e & 290 & 4.3e20 & hg22ngc4382(m85 ) & 12 25 24.7 & 18 11 27 & 25.8 & 9.95 & -22.11 & 11.01 & sa(s)0@xmath53 & 200 & 2.7e20&virgoic1459 & 22 57 10 & -36 27 42 & 43.0 & 10.83 & -22.34 & 11.10 & e3 - 4 & 316 & 1.2e20 & hg15ngc4365 & 12 24 27.9 & 07 19 06 & 25.8 & 10.42 & -21.64 & 10.82 & e3 & 262 & 1.6e20 & virgo lllrlllllllll ngc4472 & 1993/07/04 & 5 & b & 4 & 20.1 & 20.1 & 21.5 & 21.5 & 0.39 & 0.31 & 0.13 & 0.16ngc4406 & 1993/07/03 & 5 & f & 4/2 & 19.7 & 19.7 & 20.8 & 20.8 & 0.34 & 0.27 & 0.11 & 0.13ngc4636 & 1993/06/22 & 5 & b & 4 & 38.3 & 38.4 & 37.6 & 37.6 & 0.38 & 0.27 & 0.090 & 0.092ngc4649 & 1994/01/07 & 5 & f & 4/2 & 38.8 & 38.8 & 38.8 & 38.8 & 0.18 & 0.11 & 0.070 & 0.072ngc499 & 1994/01/23 & 1.4 & f & 2 & 28.0 & 28.1 & 38.9 & 38.8 & 0.031 & 0.031 & 5.1e-3 & 8.8e-3ngc507 & 1994/01/23 & 1.4 & f & 2 & 28.0 & 28.1 & 38.9 & 38.8 & 0.052 & 0.039 & 0.020 & 0.016ngc720 & 1993/07/17 & 3.3 & b & 4 & 34.9 & 35.0 & 36.3 & 36.3 & 0.033 & 0.026 & 8.7e-3 & 0.011ngc4374 & 1993/07/04 & 5 & b & 4/2 & 19.3 & 19.4 & 20.5 & 20.5 & 0.048 & 0.088 & 0.020 & 0.037ic4296 & 1994/02/15 & 1.8 & f & 2 & 35.8 & 35.8 & 37.1 & 37.1 & 0.025 & 0.018 & 8.5e-3 & 0.010ngc4382 & 1994/05/27 & 5 & f & 2/1 & 33.1 & 33.4 & 35.3 & 35.3 & 0.045 & 0.035 & 0.011 & 0.013ic1459 & 1993/05/20 & 3 & b & 4 & 19.5 & 19.5 & 19.3 & 19.3 & 0.043 & 0.032 & 0.016 & 0.018ngc4365 & 1993/06/28 & 5 & b & 4 & 35.7 & 35.7 & 36.5 & 36.5 & 0.019 & 0.019 & 0.013 & 0.015 lccccccccccccc ngc4472 & 13.4 & 0.89 & 0.39 & 5.9 & 5.5e-12 & 9.0e-13 & 6.2e41 & 8.4e40 & 442.1/425 & ( 11.315.5 ) & ( 0.880.90 ) & ( 0.330.45)&(5.26.7 ) ngc4406 & 10.9 & 0.84 & 0.31 & 6.3 & 5.6e-12 & 4.2e-13 & 6.0e41 & 3.9e40 & 708.3/616 & ( 8.513.0 ) & ( 0.830.86 ) & ( 0.260.38 ) & ( 5.27.3 ) ngc4636 & 4.9 & 0.76 & 0.31 & 5.0 & 5.9e-12 & 3.9e-13 & 5.4e41 & 3.3e40 & 732.9/428 & ( 3.66.3 ) & ( 0.750.77 ) & ( 0.280.35 ) & ( 4.45.6 ) ngc4649 & 16.1 & 0.85 & 0.34 & 2.7 & 2.3e-12 & 7.1e-13 & 2.8e41 & 6.8e40 & 605.7/545 & ( 13.019.7 ) & ( 0.830.86 ) & ( 0.280.44 ) & ( 2.23.3 ) ngc499 & 7.8 & 0.71 & 0.25 & 21.3 & 1.5e-12 & @xmath541.2e-13 & 2.0e42 & @xmath541.4e41 & 123.0/115 & ( 1.016.1 ) & ( 0.670.76 ) & ( 0.180.47 ) & ( 12.826.6 ) ngc507 & 10.7 & 1.01 & 0.29 & 33.0 & 1.9e-12 & 2.5e-13 & 2.6e42 & 3.1e41 & 317.0/177 & ( 6.4 - 15.6 ) & ( 0.961.05 ) & ( 0.200.42 ) & ( 24.542.6 ) ngc720 & 0.0 & 0.65 & 0.098 & 1.9 & 5.5e-13 & 2.2e-13 & 1.0e41 & 4.1e40 & 150.3/139 & ( @xmath541.3 ) & ( 0.610.68 ) & ( 0.0690.14 ) & ( 1.52.5 ) ngc4374 & 12.1 & 0.67 & 0.052 & 3.5 & 1.2e-12 & 4.7e-13 & 1.4e41 & 4.3e40 & 274.9/317 & ( 3.320.0 ) & ( 0.590.81 ) & ( 0.0290.12 ) & ( 1.75.5 ) ic4296 & 6.8 & 0.84 & 1.1 & 1.1 & 3.7e-13 & 4.5e-13 & 2.8e41 & 3.1e41 & 139.7/110 & ( @xmath5427.0 ) & ( 0.790.88 ) & ( @xmath550.32 ) & ( 0.183.7 ) ngc4382 & 19.0 & 0.28 & 0.056 & 4.3 & 3.7e-13 & 5.3e-13 & 7.3e40 & 5.2e40 & 280.4/217 & ( 0.4332.5 ) & ( 0.220.33 ) & ( 0.029 - 0.21 ) & ( 0.5318.3 ) ic1459 & 0.0 & 0.71 & 0.17 & 0.64 & 2.2e-13 & 7.2e-13 & 4.8e40 & 1.6e41 & 84.11/109 & ( @xmath544.1 ) & ( 0.64 - 0.79 ) & ( @xmath550.065 ) & ( 0.0241.4 ) ngc4365 & 0.0 & 0.56 & 0.12 & 0.11 & 8.4e-14 & 3.9e-13 & 6.6e39 & 3.1e40 & 200.2/218 & ( @xmath5432.8 ) & ( 0.22 - 0.72 ) & ( @xmath550.016 ) & ( 0.00228.7 )	we have systematically investigated the _ asca _ spectra of 12 early type galaxies . this paper presents the global spectral properties of these systems based on a larger sample than in any previous _ asca _ study . the x - ray spectra were uniformly fitted by a two - component model consisting of hard x - rays from thermal emission with a temperature of about 10 kev or from a power - law with index 1.8 , plus soft x - rays from a thin thermal plasma with temperature ranging from 0.3 to 1 kev . the x - ray luminosities of the hard component are found to be proportional to the blue band luminosities , while those of the soft component show large scatter with no clear correlation . the metal abundances determined from the soft component are systematically lower than solar , with a mean value of about 0.3 solar . we examine the relationships between the temperature and volume emission measure , and between the gas temperature and the stellar velocity dispersion . the volume emission measures for early type galaxies plotted as a function of the gas temperature are well below the extrapolated line found in clusters of galaxies , indicating that early type galaxies are relatively gas poor compared with galaxy clusters . the ratio of the stellar kinetic energy per unit mass to the thermal energy of the hot gas per unit mass ( @xmath0 ) is less than unity . we found no systematic relationship between x - ray properties and environment , suggesting that the interaction between interstellar matter and the intracluster medium is not strong .
optics , aberration , matrices . an optical system can be effectively modelled in paraxial approximation by a product of @xmath0 matrices , each representing one elementary transformation of the light rays @xcite ; the elementary transformations are either translations of the ray in homogeneous media or the effects of surfaces separating different media . a more accurate approach implies the consideration of higher order terms but the fact that snell s law makes use of the sine function rules out the terms of even order ; as a consequence , when one wants to improve on the paraxial approximation , one has to consider third order terms . aberrations have already been studied extensively @xcite but work is still going on in order to design symbolic models of optical systems that computers can use for optimisation purposes and humans can look at to gain a better understanding of systems performance . the matrix theory has been extended to deal with higher order terms @xcite through the use of a vector basis that incorporates two position and two orientation coordinates as well as all their third or higher order monomials , increasing the overall dimension which becomes 24 for third order approximation . it is possible to apply axis symmetry to reduce the matrix dimension through the use of complex coordinates and their higher order monomials ; for instance , in third order the matrices to be considered for axis - symmetric systems are @xmath1 @xcite . the set of four coordinates normally used to describe the ray consists of the two rectangular coordinates , @xmath2 and @xmath3 along with the two ray slopes @xmath4 and @xmath5 . i chose to replace the ray slopes with the direction cosines relative to the coordinate axes , respectively @xmath6 and @xmath7 , in order to allow an easier and more elegant formulation of the snell s law at a surface but at the expense of rendering the translation transformation non - linear . this work details a general method that can be used to establish all the coefficients needed for any order modelling of optical systems built with surfaces of unspecified shape . the power of the method is exemplified with determination of matrix coefficients for spherical systems in seventh order and for a toroidal surface in third order . the optical axis is assumed to lie along the @xmath8 axis , so the position of any point is determined by the two coordinates @xmath2 and @xmath3 ; when dealing with axis symmetric systems , the two coordinates can be combined in the complex coordinate @xmath9 . the ray orientation is defined by cosines of the angles with the @xmath2 and @xmath3 axes , respectively @xmath6 and @xmath7 ; again a complex orientation coordinate can be used , @xmath10 , in order to simplify the rotational treatment of axis symmetric systems . using the same notation of reference @xcite , @xmath11 is the vector whose components are the coordinates and their monomials up to order @xmath12 : @xmath13 where total order @xmath14 is @xmath12 or less and odd , with all exponents greater than zero . by convention the vector elements are placed in the order of smaller @xmath12 and larger four digit number @xmath15 . any transformation of @xmath11 into @xmath16 can be represented by a square matrix @xmath17 of dimension @xmath18 equal to the size of @xmath11 , following the equation : @xmath19 the matrix @xmath17 has the following structure : @xmath20 the four submatrices that form @xmath17 have the following meaning : @xmath21 contains the paraxial constants of the optical system , @xmath22 has @xmath23 high order coefficients from the series expansion of the transformed coordinates , @xmath24 is composed of zeroes and finally @xmath25 is obtained from the elements of the first four lines by a procedure called extension whereby the elements of lines 5 to @xmath18 from @xmath16 are calculated and all terms of orders higher than @xmath12 are dropped . more specifically , if we wish to determine the coefficients for the line corresponding to the monomial @xmath26 , we take the polynomial expressions for @xmath2 , @xmath3 , @xmath6 and @xmath7 from lines 1 to 4 and raise them to the powers @xmath27 , @xmath28 , @xmath29 and @xmath30 , respectively , making their product afterwards ; the result is a polynomial of degree @xmath31 from which only the terms of orders up to @xmath12 should be considered . the submatrix @xmath25 can itself be subdivided into components of different orders and components with all zero elements . although the ray transformation is described by an @xmath32 matrix , the above considerations show that only the @xmath23 elements from the first four lines need be considered , as the extension procedure is standard for all transformations . in cases where symmetry exists the number of independent coefficients can be greatly reduced @xcite . it is apparent that the matrix for any given transformation can be considered completely defined when the first four lines coefficients have been evaluated , i.e. when the transformed coordinates have been expressed in a power series of order @xmath12 . the coordinate conventions made above are not the same as those made in references @xcite and indeed by the majority of authors ; they may appear to be a poorer choice because , as we shall see , they lead to a non - linear translation transformation ; on the other hand they will simplify the determination of refraction coefficients . it should be noted , however , that a coordinate change can be seen as a transformation , itself governed by a matrix ; it is not difficult to follow the procedure outlined below and then change from cosines into slopes , if needed . the first elementary transformation that has to be considered is just the translation of the ray in an homogeneous medium ; the orientation coordinates do nt change but the position coordinates change according to the equations @xmath33 where @xmath34 is the distance travelled , measured along the optical axis . the series expansion of the equations is rather straightforward ; up to the seventh order it is : @xmath35 the previous equations give directly the coefficients for lines 1 and 2 of the matrix @xmath36 which describes the translation transformation ; lines 3 and 4 are made up of zeroes , except for the diagonal elements which are 1 ; this translates the two relations @xmath37 and @xmath38 . the lines 5 to @xmath18 can be obtained by the extension procedure described before . if slopes were used instead of direction cosines the translation matrix would have a much simpler form , corresponding to a linear transformation @xcite . the influence of the surface power on the ray coordinates must now be considered . it is useful to analyse separately the changes on the ray orientation and the modification of the position coordinates . the ray orientation changes due to the application of snell s law when the ray encounters the surface of separation between the two media ; let this surface be defined by the general formula : @xmath39 let a ray impinge on the surface on a point of coordinates @xmath2 and @xmath3 with direction cosines @xmath6 and @xmath7 . the normal to the surface at the incidence point is the vector : @xmath40 the direction of the incident ray is the unit vector : @xmath41 and the refracted ray has a direction which is represented by the vector : @xmath42 where @xmath43 and @xmath44 are the direction cosines after refraction . one can apply snell s law by equating the cross products of the ray directions with the normal on both sides of the surface multiplied by the respective refractive indices : @xmath45 where @xmath46 represents the refractive index ratio of the two media . the algebraic solutions of the equation above can be found for many surfaces by means of suitable symbolic processing software like mathematica ; in certain cases the solution is easier to find after a suitable change of coordinates along the @xmath8 axis , which will not affect the final result ; for instance , for a spherical surface it is convenient to shift the coordinate origin to the center . in order to find the coefficients for the matrix @xmath47 , representing the change in the ray orientation , the explicit expressions for @xmath43 and @xmath44 must be expanded in series ; this can again be programmed into mathematica . the examples given later show this procedure for two different surfaces . the use of just two position coordinates implies that these are referenced to a plane normal to the axis . when a ray is refracted at a surface , the position coordinates that apply are those of the incidence point ; nevertheless the translation to the surface is referenced to the vertex plane and thus the position coordinates that come out of the translation transformation , are those of the intersection with that plane ; the difference between the two positions is named surface offset . the figure [ offset ] illustrates the problem that must be solved : the reference plane for the position coordinates of the ray in a refraction process is the plane of the surface vertex ; the coordinates of the point of intersection of the ray with this plane are not the same before and after the refraction . the surface matrix must account not only for the orientation changes but also for the position offset introduced by the refraction process . if the origin of the rectangular coordinates is located at the vertex and @xmath48 are the coordinates of the ray when it crosses the plane of the vertex , the current coordinates of a point on the incident ray are given by the two equations : @xmath49 as the intersection point has to verify both ( [ eq : offset1 ] ) and ( [ eq : surface ] ) it is possible to solve a system of three simultaneous equations to determine its coordinates . in fact there may be several intersection points , as the ray may intercept a complex surface in several points ; it is not difficult , however , to select the solution of interest by careful examination of the surface region that has been hit ; we are usually interested in the solution that has the smallest @xmath50 value . as before a series expansion of the exact expressions of @xmath51 and @xmath52 must be performed taking only the terms up to the @xmath12th order ; this will yield the coefficients for the matrix @xmath53 which describes the transformation from point @xmath54 to point @xmath55 and will be called the forward offset . the offset from point @xmath55 to point @xmath56 , designated backward offset , is the reverse transformation of the forward offset , for a ray whose direction coordinates have been modified by refraction . we solve ( [ eq : offset1 ] ) and ( [ eq : surface ] ) in terms of @xmath2 and @xmath3 and apply the previous procedure to determine the coefficients of the transformation matrix , @xmath57 . the matrix describing the transformations imposed by the surface is the product @xmath58 , as the ray has to suffer the three successive transformations : forward offset , refraction and backward offset . the surface matrix is called @xmath59 . the entrance pupil of the optical system plays an important role in the overall aberrations ; the center of the pupil defines the chief ray , which determines the paraxial image point . the ray fan usually extends equally in all directions around the chief ray and the points of intersection of the various rays in the fan with the image plane determine the ray aberrations . when matrices are used to model the system the image appears described in terms of the position and orientation coordinates of the rays when these are subjected to the first transformation , be it a refraction or a translation ; in terms of ray aberrations this corresponds to a situation where the object is located at infinity and the entrance pupil is placed where the first transformation takes place . in fact , the orientation coordinates play the role of object coordinates and the position coordinates are the actual pupil coordinates , if this coincides with the first transformation ; the chief ray can be found easily just by zeroing the position coordinate . ray aberrations can be evaluated correctly if the image is described in terms of both object and stop coordinates or some appropriate substitutes ; an object point is then set by the object coordinates , the paraxial image point is found when the stop coordinate is zero and the aberrations are described by the differences to this point . this can be done adequately when the stop is located before the first surface and so coincides with the entrance pupil . if the object is at infinity the problem is very easy to solve and application examples have already been described @xcite . if the entrance pupil is located at the position @xmath60 relative to the first surface , this means that the rays must perform a translation of @xmath61 prior to hitting it ; this can be accommodated via the right product by a translation matrix @xmath62 , where the index indicates the amount of translation . [ htb ] the case of near objects is more difficult to deal with because we are faced with a dilemma : if a translation from the pupil position to the first surface is applied the position coordinates of the incident rays become pupil coordinates but the rays origins on the object are lost . conversely we can apply a translation from the object position , ending up with the reverse problem of knowing the rays origins and ignoring their point of passage through the pupil . we need the position coordinates both at the object and pupil location ; this suggests that the ray s orientation should be specified not by its direction cosines but by the coordinates of the point of passage through the pupil . in figure [ pupil ] a ray leaves an object point of coordinates @xmath63 and crosses the entrance pupil plane at a point of coordinates @xmath64 ; then the direction cosines , @xmath6 and @xmath7 can be calculated by : @xmath65 where @xmath66 is the position of the pupil relative to the object . the matrix theory developed before was based on direction cosines rather than pupil coordinates ; in consequence a coordinate change is needed before the translation from the object to the first surface is applied to the rays . rather than use equations ( [ eq : pupil ] ) it is more convenient to look at them as a coordinate change governed by a matrix @xmath67 ; for the determination of its coefficients we apply the standard procedure of series expansion . the following is the result for 7th order : @xmath68 a similar expression exists for @xmath7 . the equation above allows the determination of the coefficients for matrix @xmath67 which converts the coordinates @xmath69 into @xmath70 before the rays enter the system at the object position ; as a symmetrical conversion is not performed when the rays leave the system , there is no place for the use of matrix @xmath71 , as would be the case in a similarity transformation . a case study would involve the following successive steps : * determine the matrices for all the translation and refraction transformations suffered by the ray from the object point to the image plane . * determine the matrix @xmath67 , relative to the entrance pupil position . * multiply all the matrices in reverse order , starting with the translation from the last surface to the image plane and ending with matrix @xmath67 ; the result is matrix @xmath17 . * determine a vector base @xmath11 with the set of coordinates @xmath72 . * make the product @xmath73 . once @xmath16 has been calculated , its first four elements are polynomials of @xmath12th degree on the independent variables @xmath72 , which model the position and direction cosines of the rays on the image plane . most optical systems have stops located after the first surface , and so the method described above is not adequate . one can always find the gaussian entrance pupil of any system and so revert to the previous situation ; this will work even if the entrance pupil is found to lie beyond the first surface , in which case it will give rise to a negative translation . the problem is that the gaussian entrance pupil is a first order transformation of the stop , whereas the real entrance pupil is an aberrated image of the stop given by the portion of the optical system that lies before it . ideally one should try to describe the image in terms of the object coordinates and the point of passage through the stop , which can be rather difficult . alternatively one can find a reference ray , defined as the ray that passes through the center of the stop and is not necessarily the same as the chief ray , and work with position and orientation coordinates on the image plane , relative to the reference ray . the plot of the relative ray position versus its relative direction cosines can be used to describe ray aberrations @xcite . a sphere of radius @xmath74 with its center at @xmath75 is defined by the equation : @xmath76 this equation will replace ( [ eq : surface ] ) in the algorithm for the determination of matrix coefficients . in a previous work @xcite we presented the results for third order ; these are now extended to seventh order . this is a situation where symmetry must be considered in order to reduce the size of the matrices involved ; we use the complex variables @xmath54 and @xmath77 previously defined , leading to a complex vector base @xmath78 whose elements have the general form @xmath79 with @xmath80 and @xmath81 representing the complex conjugates of @xmath54 and @xmath77 . kondo et al . @xcite have shown that the powers must obey the condition @xmath82 ; for seventh order the allowable combinations are : @xmath83 equation ( [ eq : snell ] ) can now be solved and the resulting expressions for @xmath43 and @xmath44 can be expanded in series up to the seventh order ; the results can be combined into one complex variable @xmath84 . the second column of table [ t : spherical ] lists the coefficients for this expansion . similarly the forward offset coefficients can be found solving equations ( [ eq : offset1 ] ) and combining @xmath51 and @xmath52 expansions into one single complex variable @xmath55 ; the resulting coefficients are listed in column 3 of table [ t : spherical ] . the backward offset coefficients are obtained in a similar way , although now it is the @xmath2 and @xmath3 expansions that must be combined into a single complex variable @xmath56 . toroidal surfaces are used to correct eye astigmatism ; they can be fabricated easily without resorting to sophisticated machinery . in this example we deem to show that the method is applicable to general surfaces but it is not our purpose to list high order coefficients for this particular shape ; therefore we will restrict the study to third order . also , as there is no rotational symmetry we will not use complex coordinates ; there is symmetry relative to two perpendicular planes and if we were to use complex coordinates the terms should obey the condition @xmath85 as explained in reference @xcite ; we feel that it is clearer not to invoke symmetry at all . the surface is generated by a circle of radius @xmath86 normal to the @xmath87 plane whose center is displaced along a circle of radius @xmath88 laying on this plane . this surface has curvature radii of @xmath86 and @xmath89 respectively on the @xmath87 and @xmath90 planes . the surface equation can be written as : @xmath91 the solution of equation ( [ eq : snell ] ) originates the expressions for @xmath43 and @xmath44 ; these are then expanded in series up to the third order and the resulting coefficients are listed in columns 2 and 3 of table [ t : toroid ] . the same procedure was applied to the expressions for @xmath51 and @xmath52 resulting from equation ( [ eq : offset1 ] ) ; columns 4 and 5 of the same table list these coefficients . the backward offset coefficients need not be listed , as in third order they can be obtained by sign reversal from the forward offset coefficients . matrices can be used to model optical systems built with surfaces of unspecified shape to any desired degree of approximation . the method that was described differs from those found in literature by the choice of direction cosines rather than ray slopes to specify ray orientations . this results in a very elegant formulation of snell s law which can be easily programmed into symbolic computation software . in spite of some added complication in translation matrices the algorithm is limited only by computing power in it s ability to determine the matrix coefficients , irrespective of surface complexity and order of approximation . for axis - symmetric systems the matrices have a dimension @xmath92 for seventh order and can usually be dealt with by ordinary pcs ; the influence of stops can also be incorporated in the matrix description as was demonstrated . the unconventional coordinate system that was used can , if needed , be converted to the more usual ray slope coordinate system through the product with conversion matrices of the same dimension . two examples were shown for illustration of the method s capabilities ; the first one supplies a list of seventh order coefficients for systems made up with spherical surfaces and allows the reader an immediate use . the second example applies the method to a fourth degree toroidal surface , showing that complex shapes can be accommodated . almeida , `` the use of matrices for third order modeling of optical systems '' , in _ international optical design conference _ , k.p . thompson and l.r . gardner , eds . spie * 3482 * , 917 - 925 , ( 1998 ) .	the non - linear transformations incurred by the rays in an optical system can be suitably described by matrices to any desired order of approximation . in systems composed of uniform refractive index elements , each individual ray refraction or translation has an associated matrix and a succession of transformations correspond to the product of the respective matrices . this paper describes a general method to find the matrix coefficients for translation and surface refraction irrespective of the surface shape or the order of approximation . the choice of coordinates is unusual as the orientation of the ray is characterised by the direction cosines , rather than slopes ; this is shown to greatly simplify and generalise coefficient calculation . two examples are shown in order to demonstrate the power of the method : the first is the determination of seventh order coefficients for spherical surfaces and the second is the determination of third order coefficients for a toroidal surface .
in trujillo & sheppard ( 2014 ) , the extreme trans - neptunian objects or etnos are defined as asteroids with heliocentric semimajor axis greater than 150 au and perihelion greater than 30 au . there are at present 16 known etnos ( see tables [ helio ] and [ bary ] for relevant data ) ; they exhibit clustering in the values of their argument of perihelion ( trujillo & sheppard 2014 ) , longitude of the ascending node ( batygin & brown 2016 ) , eccentricity and inclination ( de la fuente marcos & de la fuente marcos 2014 , 2016 ) . the analysis in de la fuente marcos & de la fuente marcos ( 2014 ) has shown that the clustering in eccentricity ( @xmath1 about 0.81 ) can be due to a selection effect , but this can not be the case of the ones found in inclination ( @xmath2 about 20 ) , longitude of the ascending node ( @xmath3 about 134 ) , and argument of perihelion ( @xmath4 about @xmath526 ) . these patterns could be induced by the gravitational perturbation from one ( trujillo & sheppard 2014 ; gomes , soares & brasser 2015 ; batygin & brown 2016 ) or more ( de la fuente marcos & de la fuente marcos 2014 ; de la fuente marcos , de la fuente marcos & aarseth 2015 ) , yet to be discovered , trans - plutonian planets . malhotra , volk & wang ( 2016 ) have pointed out that the four etnos with the longest orbital periods ( see table [ bary ] ) have simple numerical relationships between periods . this implies that the etnos could be a highly structured , dynamically speaking , population . the etnos are not massive enough to induce orbit - orbit coupling by themselves , but a hypothetical planet or planets could be the source of these accidental commensurabilities . malhotra et al . ( 2016 ) used their analysis to predict the existence of a planet with semimajor axis @xmath0665 au that is only marginally compatible with the one discussed by batygin & brown ( 2016 ) within the framework of their planet nine hypothesis ; even if their proposed value for the semimajor axis is similar , @xmath0665 au versus @xmath0700 au , the preferred values of @xmath2 and @xmath3 in malhotra et al . ( 2016 ) (@xmath6 ) or ( @xmath7) are rather different from those in batygin & brown ( 2016 ) (@xmath8 ) . accidental commensurabilities are a natural by - product of the existence of mean motion resonances in the solar system ( for the theory of orbital resonance , see e.g. murray & dermott 1999 ) . the orbital architecture of the main asteroid belt is mainly the result of interior mean motion resonances with jupiter ( see e.g. holman & murray 1996 ; nesvorn & morbidelli 1999 ) ; conversely , the one observed in the trans - neptunian belt is induced by exterior mean motion resonances with neptune ( see e.g. gladman et al . 2012 ) . in the main asteroid belt , the dominant ones are located at 2.06 au ( 4:1 ) , 2.5 au ( 3:1 ) , 2.82 au ( 5:2 ) , 2.95 au ( 7:3 ) and 3.27 au ( 2:1 ) ; therefore , any asteroids trapped in these resonances are also nominally resonant among themselves even if these objects are not massive enough to induce orbit - orbit coupling by themselves . for instance , the alinda asteroids which are subjected to the 3:1 mean motion resonance with jupiter and the griqua asteroids that occupy the 2:1 resonance are also in a mutual 3:2 near mean motion resonance . for objects trapped in the dominant resonances , the values of the ratios of orbital periods are : @xmath9 this sequence represents a characteristic sample of accidental commensurabilities in the main asteroid belt . in the trans - neptunian belt , the dominant mean motion resonances with neptune are located at 39.4 au ( 2:3 ) , 42.3 au ( 3:5 ) , 43.7 au ( 4:7 ) , 47.8 au ( 1:2 ) and 55.4 au ( 2:5 ) ; the associated ratios of orbital periods are : @xmath10 they define accidental commensurabilities beyond neptune . out of 10 values , as many as five could be common to both sequences . the commensurability analysis carried out in malhotra et al . ( 2016 ) provides some support for the planet nine hypothesis and singles out some resonant orbital relations , but it does not explore the subject of commensurabilities between etnos in depth . here , we use the seminal idea presented in malhotra et al . ( 2016 ) and the fact that very eccentric orbits are affected by mean motion resonances no matter the value of the semimajor axis ( see e.g. gomes et al . 2005 ) to investigate this subject extensively . we use their heliocentric and barycentric semimajor axes , their uncertainties , and monte carlo techniques to find out if the values of the ratios of orbital periods of the etnos match those in the main and trans - neptunian belts . this letter is organized as follows . section 2 presents the data used in this study . our monte carlo methodology is explained and applied in section 3 . results are discussed in section 4 and conclusions are summarized in section 5 . ignoring the mass of the object , expressing distances in astronomical units , times in sidereal years , and masses in solar masses , kepler s third law applied to an object orbiting the sun states that @xmath11 , where @xmath12 is the period and @xmath13 is the semimajor axis of the orbit of the object . given two objects the first one with @xmath14 and @xmath15 , and the second one with @xmath16 and @xmath17 , @xmath18 kepler s third law is given by @xmath19 . if @xmath20 can be written in the form of a ratio of small integers then we have a mean motion resonance . in order to investigate the distribution of ratios of orbital periods , @xmath20 , for the etnos the values of the semimajor axes are needed . the orbital solutions of the etnos are far less robust than those available for the best studied members of the main asteroid belt or the trans - neptunian belt ; the values of the relative error in semimajor axis are several orders of magnitude larger in the case of the etnos . any study of commensurabilities between etnos must account for this fact ; a probabilistic approach based on the values of their uncertainties can be used to address this important issue . the values of the heliocentric semimajor axes and their uncertainties from the jet propulsion laboratory , jpl , small - body database and horizons on - line ephemeris system ( giorgini et al . 1996 ) for the 16 objects discussed in this letter are given in table [ helio ] . three objects 2003 ss@xmath21 , 2010 gb@xmath22 and 2013 rf@xmath23 have relative uncertainties in @xmath13 greater than five per cent as their orbits are still in need of further improvement . however , the heliocentric orbital elements may not be adequate for this study as the objects discussed here are too distant from the sun . in cases like this one , the use of barycentric orbital elements is perhaps more appropriate ( todorovic - juchnicwicz 1981 ; malhotra et al . 2016 ) . 0.35truecm .heliocentric semimajor axes and their uncertainties for the 16 objects discussed in this letter . ( epoch : 2457400.5 , 2016-january-13.0 00:00:00.0 tdb . j2000.0 ecliptic and equinox . source : jpl small - body database . data retrieved on 2016 march 13 . ) [ cols="^,^",options="header " , ] [ bary ] in this section we construct commensurability maps by generating pairs of virtual etnos from the data in tables [ helio ] and [ bary ] via monte carlo techniques ( metropolis & ulam 1949 ; press et al . 2007 ) and evaluating their commensurability parameter or ratio of orbital periods , @xmath24 , enforcing that @xmath25 by swapping data if necessary . the value of the semimajor axis of a virtual etno is computed using the expression @xmath26 , where @xmath27 is the semimajor axis of the virtual etno , @xmath28 is the mean value of the semimajor axis from the available orbit ( tables [ helio ] and [ bary ] ) , @xmath29 is the standard deviation of @xmath13 ( table [ helio ] ) , and @xmath30 is a ( pseudo ) random number with normal distribution . in our calculations , the box - muller method ( box & muller 1958 ; press et al . 2007 ) was used to generate random numbers from the standard normal distribution with mean 0 and standard deviation 1 . when computers are used to produce a uniform random variable ( to seed the box - muller method ) it will inevitably have some inaccuracies because there is a lower bound on how close numbers can be to 0 . for a 64 bits computer the smallest non - zero number is @xmath31 which means that the box - muller method will not produce random variables more than 9.42 standard deviations from the mean ; however , unphysical values have not been used in the calculations . the number of pairs of virtual etnos tested per monte carlo experiment is @xmath32 . a random ( uniform ) distribution of semimajor axes should exhibit a degree of commensurability compatible with zero . we have generated ten sets of 16 values of the semimajor axis in the range that appears in table [ helio ] or [ bary ] and assigned the uncertainties in table [ helio ] to them . we have obtained the frequency distribution of commensurability parameters for each set and computed the average frequency distribution . the results appear in purple in figs [ commhelio ] and [ commbary ] . however , the etnos have been discovered from the ground and this fact induces a bias in the observed distribution of semimajor axes even if the etnos are not subjected to any external perturbation and just follow heliocentric , or barycentric , orbits . following the analysis in de la fuente marcos & de la fuente marcos ( 2014 ) and , in order to reproduce this bias , we generate ten sets of 16 values of the semimajor axis according to the distribution in the top panel of fig . 3 in de la fuente marcos & de la fuente marcos ( 2014 ) ; i.e. objects with perigee @xmath33 au and declination at perigee in the range ( @xmath34 ) , see fig . 2 in de la fuente marcos & de la fuente marcos ( 2014 ) . the average frequency distribution of this biased sample appears in green in figs [ commhelio ] and [ commbary ] . the error bars have been computed assuming poisson statistics , @xmath35 , and using the approximation given by gehrels ( 1986 ) when @xmath36 : @xmath37 , where @xmath38 is the number of pairs . in both cases the results are compatible with a null degree of commensurability . any statistically significant deviation from this null result should be interpreted as resulting from a present - day gravitational perturbation located within the region travelled by the etnos . using the data in table [ helio ] , we created @xmath39 pairs of virtual etnos following the procedure outlined above and computed the commensurability parameter . the commensurability map in terms of colour is shown in fig . [ commhelio ] ( the value of the commensurability parameter or ratio of orbital periods is plotted in colour , top panel ) , where each point represents a pair of virtual etnos ; the associated frequency distribution is plotted in fig . [ commhelio ] ( bottom panel ) . there are obvious and statistically significant deviations from the uniform or biased expectations explored above . the most conspicuous commensurabilities in fig . [ commhelio ] ( bottom panel ) are 1.0 , 1.11 , 1.65 and 1.8 . . frequency distribution ( bottom panel ) in @xmath24 from the commensurability map . the results from a uniform spread in semimajor axis are plotted in purple and those from a biased uniform sample in green . the number of bins in the frequency distribution plot is 2 @xmath40 , where @xmath38 is the number of pairs of virtual etnos tested , @xmath32 . the black curve shows the cumulative distribution . ] using the data in table [ bary ] , we have obtained fig . [ commbary ] . as for the heliocentric orbits , there are significant deviations from the unperturbed scenario . considering the values of the barycentric semimajor axes , the most obvious commensurabilities in fig . [ commbary ] are 1.02 , 1.11 , 1.38 , 1.65 and 1.71 . three of the statistically significant commensurabilities coincide with those found in the analysis of heliocentric orbits . in addition , several of these values are also present in the sequences associated with the main and trans - neptunian belts . this is unlikely to be the result of chance alone . but for the barycentric values in table [ bary ] . ] the analysis presented in the previous section shows that the frequency distribution of the ratio of orbital periods of known etnos is statistically incompatible with that of an unperturbed asteroid population following heliocentric / barycentric orbits . this result is unlikely to be a statistical artefact as the known etnos have been discovered by several independent surveys with , presumably , uncorrelated biases . if the frequency distribution of the ratio of periods is incompatible with an unperturbed scenario , how significant are the favoured commensurabilities ? in order to answer this rather critical question , we have computed the difference between the frequency obtained from the observational values and that from the biased expectation ( green curve in figs [ commhelio ] and [ commbary ] ) and divided by the value of the standard deviation in the biased case calculated as described above . our results are plotted in fig . [ sigma ] . the number of statistically significant commensurabilities is larger than the few indicated above and found by simple visual inspection . the outcome is not dependent on the bin size ; we have experimented with @xmath41 and @xmath42 and the results are fully consistent . the statistical significance analysis in fig . [ sigma ] shows that our results are robust . probabilistically speaking , there are simply too many etnos in commensurabilities compatible with the action of massive perturbers . regarding the extremely significant value of the ratio of periods @xmath43 in fig . [ sigma ] , we would like to downplay this issue as two objects have values of the semimajor axis of @xmath44 au and several others have semimajor axes close to 150 au . these two values are the abrupt cutoffs in the distribution of semimajor axes . of the deviations of the frequency distribution in the heliocentric ( blue ) and barycentric ( red ) cases ( data from figs [ commhelio ] and [ commbary ] ) with respect to the biased expectations . the @xmath45 limits are also indicated . ] one of the most surprising properties of the sample studied is the unusually large fraction of objects with ratio of periods @xmath46 . this issue was already noticed in de la fuente marcos & de la fuente marcos ( 2014 ) . in particular , the orbits of 2004 vn@xmath47 and 2013 rf@xmath23 are alike , and several objects have values of the barycentric semimajor axis within narrow ranges of each other . this could mean that some of these objects move co - orbital , i.e in a 1:1 mean motion resonance , with an unseen planet ( or , more likely , several of them ) . co - orbital motion is possible at high eccentricity ( namouni 1999 ; namouni , christou & murray 1999 ; namouni & murray 2000 ) . however , a more likely scenario is in the presence of higher order resonances like the 7:3 and 5:2 in the main asteroid belt that give a ratio of periods of 1.0699 or the 4:7 and 3:5 in the trans - neptunian belt that give a ratio of 1.0501 . interior and exterior resonances are possible if there are multiple trans - plutonian planets . regarding the impact of this resonant scenario on the planet nine hypothesis , it is clearly compatible with it . if the mechanisms responsible for inducing dynamical structure in the main and trans - neptunian belts are also at work in the region occupied by the etnos then we should expect that for two given etnos in near commensurability : @xmath48 where @xmath49 is the semimajor axis of the orbit of the perturber and @xmath50 are all small integers . this expression can be applied to two of the most clear clusterings in semimajor axis in table [ bary ] . the average value of the barycentric semimajor axis of ( 90377 ) sedna and 2007 tg@xmath21 is 504 au . on the other hand , the equivalent mean value for 2004 vn@xmath47 , 2010 gb@xmath22 and 2013 rf@xmath23 is 332 au . these five objects are part of the set of six singled out by batygin & brown ( 2016 ) . the associated period ratio for these two sets of etnos is 1.87 . in the main asteroid belt , this ratio is obtained for objects trapped in the 5:3 mean motion resonance with jupiter and those in the 3:1 , that is one of the main resonances in the outer belt ( holman & murray 1996 ) . making a dynamical analogy between the two situations and decomposing eqn . [ key ] in two we have : @xmath51 and @xmath52 . the average of the two values of @xmath49 is @xmath0700 au which is the favoured value for the semimajor axis of planet nine in batygin & brown ( 2016 ) . the 1.8 commensurability has a statistical significance of 239@xmath53 for heliocentric orbits , the 1.89 commensurability has 51@xmath53 for barycentric orbits ( see fig . [ sigma ] ) . this is unlikely to be mere coincidence . another example of the potential implications of our findings arises when we focus on 2003 hb@xmath54 , 2015 so@xmath55 , 2005 rh@xmath56 , ( 445473 ) 2010 vz@xmath23 and 2013 gp@xmath57 , the first two could be in a 3:2 resonance with a hypothetical planet at @xmath58 au , with the other three in a 5:3 resonance with the same planet . in this framework , the pair 2003 hb@xmath54 and 2015 so@xmath55 would be in a 10:9 accidental resonance with the other three etnos . a ratio of periods @xmath59 is present in both the main asteroid belt and the trans - neptunian belt . the 1.1 commensurability has a statistical significance of 76@xmath53 for heliocentric orbits and 61@xmath53 for barycentric orbits ( see fig . [ sigma ] ) . on the other hand , the 1.65 commensurability is present for both heliocentric ( 192@xmath53 ) and barycentric ( 131@xmath53 ) orbits ; a similar analysis focusing on 2003 hb@xmath54 , 2013 gp@xmath57 , ( 82158 ) 2001 fp@xmath60 and 2002 gb@xmath61 is compatible with a hypothetical planet at @xmath62 au considering resonances ( 3/1)(5/9)=5/[email protected] . with the currently available data , degenerate solutions are possible , but they still hint at a multi - planet scenario . trans - plutonian planets may have been scattered out of the region of the giant planets early in the history of the solar system ( see e.g. bromley & kenyon 2014 , 2016 ) , but planets similar to uranus or neptune may also form at 125750 au from the sun ( kenyon & bromley 2015 , 2016 ) . in this letter , we have explored the existence of commensurabilities between the known etnos . this analysis has been inspired by the hypothesis and discussion presented in malhotra et al . given the fact that these objects are not massive enough to induce orbit - orbit coupling by themselves , the existence of statistically significant commensurabilities could only signal the presence of massive unseen perturbers . summarizing , our conclusions are as following . * the clustering in orbital parameter space observed for the heliocentric orbits of the known etnos is also present when considering their barycentric orbits . in particular , @xmath1 clumps about [email protected] , @xmath2 about 20@xmath638 , @xmath3 about 134@xmath6372 , and @xmath4 about @xmath526@xmath6349 . statistical outliers do exist . * the frequency distribution of the ratio of orbital periods of known etnos is statistically incompatible with that of an unperturbed asteroid population following heliocentric / barycentric orbits . in contrast , it resembles the ones present in the main and trans - neptunian belts . the existence of statistically significant accidental commensurabilities between etnos strongly suggests that external perturbers induce the observed dynamical structure . * the fraction of known etnos in 1:1 commensurabilities is dozens of times the expected value for a random population . * a number of known etnos may be trapped in the 5:3 and 3:1 mean motion resonances with a putative planet nine with semimajor axis @xmath0700 au . we must stress that our results are based on small number statistics , but it is also true that the deviations from what is expected for an unperturbed asteroid population following heliocentric / barycentric orbits are so strong that what is observed is unlikely to be the result of chance alone . gravitational perturbations with sources in the region travelled by the etnos are the most probable explanation for the observed patterns . we thank the anonymous referee for her / his constructive and helpful report , and s. j. aarseth , d. p. whitmire , g. carraro , d. fabrycky , a. v. tutukov , s. mashchenko , s. deen and j. higley for comments on etnos and trans - plutonian planets . this work was partially supported by the spanish ` comunidad de madrid ' under grant cam s2009/esp-1496 . in preparation of this letter , we made use of the nasa astrophysics data system , the astro - ph e - print server , and the mpc data server . 99 batygin k. , brown m. e. , 2016 , aj , 151 , 22 box g. e. p. , muller m. e. , 1958 , annals math . stat . , 29 , 610 bromley b. c. , kenyon s. j. , 2014 , apj , 796 , 141 bromley b. c. , kenyon s. j. , 2016 , apj , submitted ( arxiv:1603.08010 ) de la fuente marcos c. , de la fuente marcos r. , 2014 , mnras , 443 , l59 de la fuente marcos c. , de la fuente marcos r. , 2016 , mnras , 459 , l66 de la fuente marcos c. , de la fuente marcos r. , aarseth s. j. , 2015 , mnras , 446 , 1867 gehrels n. , 1986 , apj , 303 , 336 giorgini j. d. et al . , 1996 , baas , 28 , 1158 gladman b. et al . , 2012 , aj , 144 , 23 gomes r. s. , gallardo t. , fernndez j. a. , brunini a. , 2005 , celest . mech . dyn . , 91 , 109 gomes r. s. , soares j. s. , brasser r. , 2015 , icarus , 258 , 37 holman m. j. , murray n. w. , 1996 , aj , 112 , 1278 kenyon s. j. , bromley b. c. , 2015 , apj , 806 , 42 kenyon s. j. , bromley b. c. , 2016 , apj , submitted ( arxiv:1603.08008 ) malhotra r. , volk k. , wang x. , 2016 , apj , submitted ( arxiv:1603.02196 ) metropolis n. , ulam s. , 1949 , j. am . , 44 , 335 murray c. d. , dermott s. f. , 1999 , solar system dynamics . cambridge univ . press , cambridge , p. 97 namouni f. , 1999 , icarus , 137 , 293 namouni f. , murray c. d. , 2000 , celest . astron . , 76 , 131 namouni f. , christou a. a. , murray c. d. , 1999 , phys lett . , 83 , 2506 nesvorn d. , morbidelli a. , 1999 , in henrard j. , ferraz - mello s. , eds , proc . iau colloq . 172 , impact of modern dynamics in astronomy . kluwer academic publishers , dordrecht , p. 381 press w. h. , teukolsky s. a. , vetterling w. t. , flannery b. p. , 2007 , numerical recipes : the art of scientific computing , 3rd edn . cambridge univ . press , cambridge todorovic - juchnicwicz b. , 1981 , acta astron . , 31 , 191 trujillo c. a. , sheppard s. s. , 2014 , nature , 507 , 471	many asteroids in the main and trans - neptunian belts are trapped in mean motion resonances with jupiter and neptune , respectively . as a side effect , they experience accidental commensurabilities among themselves . these commensurabilities define characteristic patterns that can be used to trace the source of the observed resonant behaviour . here , we explore systematically the existence of commensurabilities between the known etnos using their heliocentric and barycentric semimajor axes , their uncertainties , and monte carlo techniques . we find that the commensurability patterns present in the known etno population resemble those found in the main and trans - neptunian belts . although based on small number statistics , such patterns can only be properly explained if most , if not all , of the known etnos are subjected to the resonant gravitational perturbations of yet undetected trans - plutonian planets . we show explicitly that some of the statistically significant commensurabilities are compatible with the planet nine hypothesis ; in particular , a number of objects may be trapped in the 5:3 and 3:1 mean motion resonances with a putative planet nine with semimajor axis @xmath0700 au . [ firstpage ] methods : statistical celestial mechanics kuiper belt : general minor planets , asteroids : general oort cloud planets and satellites : general .
after the discovery of standard model ( sm ) higgs - like boson , the next step is to measure its couplings precisely to see possible deviation from the sm and thus to search for new physics . the yukawa couplings between higgs boson and charged leptons that cause lepton flavor violation ( lfv ) are of particular interest , because in the sm they are absent at tree level and highly suppressed at loop levels , and thus are sensitive to new physics . as a matter of fact , in most of models that address neutrino masses and oscillations , lfv is well expected and has already been observed in neutrino oscillations described by the pmns matrix . moreover , although charged lepton flavor violation ( clfv ) has not been observed yet , in general those models should leave measurable signals in processes like @xmath2 , @xmath3 , etc . a lot of efforts have been devoted to searching for clfv and the null results impose very strong bounds on the magnitude of lfv @xcite . searching for lfv higgs decays @xcite receives special attention in the lhc era @xcite . the cms collaboration reported the upper limit br@xmath4 at 95@xmath5 c.l . , using the 19.7 fb@xmath6 of @xmath7 tev data @xcite . interestingly , the best fit ( assuming both the production cross section and total width of higgs being sm - like ) hints a 2.4 @xmath8 excess with br@xmath9 . more recently , the atlas collaboration obtained an the upper limit br@xmath10 from hadronic @xmath11 decay at 95@xmath5 c.l . , using the 20.3 fb@xmath6 of @xmath7 tev data @xcite . although they have not seen significant deviation from the sm , their best fit value br@xmath12 is consistent with the cms result . at the 300 fb@xmath6 of 13 tev lhc , the sensitivity can reach down to 7.7@xmath13 and thus the cms excess will be confirmed or excluded @xcite . in the models with canonical seesaw mechanism lfv higgs decay is too small to be observed @xcite . this is because of the decoupling of right - handed neutrinos ( rhns ) either through the smallness of yukawa couplings or heaviness of rhns . in the inverse seesaw mechanism , where sizable yukawa couplings are allowed for light rhns , appreciable lfv higgs decay can be accommodated @xcite . alternatively , the tiny neutrino masses can be generated by radiative corrections @xcite . however , to our knowledge , none of those radiative seesaw models could generate large lfv higgs decay . actually , facing the stringent constraint from clfv , it is quite nontrivial to get lfv higgs decay large enough to detect at the lhc . at tree level , two ( or even more)-higgs doublet model ( 2hdm ) with proper flavor changing neutral current allows lfv higgs decay which is large enough to explain the cms excess @xcite . higher dimensional operators in the effective theory framework were also considered @xcite . but at loop level a large cancellation probably is needed to evade the clfv constraint @xcite . other scenarios can be found in ref . @xcite . in this article we establish a connection between lfv higgs decay and a type of dark matter ( dm ) , _ i.e. _ , lepton - flavored dm @xcite . in this scenario , dm interacts merely with the sm lepton sector , whereupon dm - quark interactions arise at loop level . an obvious merit of that kind of dm is that we can easily understand the null results from dm direct detection experiments such as lux @xcite . that paradigm can be achieved in two ways . one way is introducing a leptophilic vector boson or higgs boson propagating in the @xmath14channel for the dm pair annihilation diagrams . this kind of model gives rise to poor flavor phenomenology . the other way is introducing mediators in the @xmath15channel to form lepton flavored dm . then , lfv can happen in the dark sector and is mediated to the sm sector via loop processes . furthermore , mediators could consist of both left - handed and right - handed scalar leptons ( the previous studies were based on only one type of them ) , just as in the case of the supersymmetric sms . remarkably , we find that this kind of lepton - flavored dm is able to accommodate lfv higgs decay while other models with only one type chirality fail to . as an example , we will show that in our model a sizable br@xmath16 at the level of @xmath17 can be naturally achieved without incurring too large br(@xmath18 ) . it is attributed partly to the large chirality flipping in the scalar sector and also to the cancellation between different contributions to clfv . in addition , we study the mechanism for dm , a majorana fermion , to acquire correct relic density . for the weak scale dm , even @xmath14wave annihilation may work without large yukawa couplings . related to radiative lfv higgs decay , radiative correction could also lead to higgs - mediated dm - nucleon scattering which may be detected in the near future . this paper is organized as follows : in section [ sec : model ] the model is introduced . in section [ sec : lfv ] we consider higgs lfv decay confronting charged lepton lfv decay , along with others . in section [ sec : dm ] we study relic density and direct detection of our leptophilic dark matter and their relations with lfv higgs decay . we conclude in section [ sec : concl ] . in this section we will first present the model in its simplest version , and then calculate the mass spectra that will be used later . from model building perspective , a natural way to realize a lepton - flavored dm is to introduce a majorana dm candidate connected to some lepton flavors by means of scalar leptons . if dm is a scalar field , whether it is real or complex , it is hard to get rid of the conventional higgs portal in a natural manner , not to mention other demerits . so we focus on the case where dm is a singlet majorana fermion @xmath19 , protected by a @xmath20 dark matter parity . at the renormalizable level , @xmath19 can not couple to sm fields . its interactions with sm fields necessitate additional mediators , and we can specify these interactions by introducing mediators with proper quantum numbers . in order to make up a lepton - flavored dm , one can designate a scalar partner for each sm left - handed lepton doublet @xmath21 and right - handed lepton signlet @xmath22 . they are labelled as @xmath23 and @xmath24 , respectively . for simplicity , only a single family of scalar lepton ( slepton for short , borrowing the name from supersymmetry ) will be considered . in this paper we do not have the ambition to address the flavor structure of the dark sector by imposing flavor symmetry . we just treat all the couplings as free parameters . with the degrees of freedom at hand , restricted by the @xmath20 dark matter parity under which only the new particles are odd , the most general lagrangian ( aside from the kinetic energy terms ) takes a form of @xmath25 where @xmath26 . in our convention @xmath23 is assigned with the same hypercharge @xmath27 with the sm higgs doublet @xmath28 so that @xmath23 can be regarded as the 2nd higgs doublet in 2hdm . couplings @xmath29 and @xmath30 are not important in our ensuing discussions and are set to be zero . the part involving the two higgs doublets , as usual , is given by @xmath31 in this potential most parameters are irrelevant to our phenomenological studies , except for @xmath32 that is crucial in neutrino mass generation . a comment deserves special attention . we start from lepton - flavored dm , but as a bonus nonzero neutrino masses are generated as a generic consequence of this type of dm model . it is obvious that all of the crucial ingredients of the ma s model @xcite are incorporated in our framework , and thus radiative corrections lead to neutrino masses : @xmath33 with @xmath34 gev . in the parameter space relevant to lfv higgs decay , @xmath35 is around the weak scale while @xmath36 and moreover @xmath37 . then the resulting neutrino mass scale is much above the ev scale except for extremely suppressed @xmath38 . in this paper we will not pay further attention on this aspect and always assume a sufficiently small @xmath32 to suppress radiative neutrino mass . in the right vacuum , only @xmath28 is supposed to develop vacuum expectation value ( vev ) , breaking the electroweak symmetries but not @xmath20 . then the charged component of @xmath23 , which is written in component as @xmath39 , would mix with @xmath24 through the @xmath40term , i.e. , @xmath41 . then mass eigenstates are related to the flavor eigenstates via [ ] e_1=(_^+)^- _ e , e_2=(_^+)^+ _ e , the two charged sleptons respectively have the following ( mass)@xmath42 [ ] m_e_1,2 ^ 2= , respectively . the @xmath43term contributions to masses have been absorbed into the bare mass term of @xmath23 , @xmath44 which is common to all components . and similar operation is done for @xmath24 . the mixing angle , within @xmath45 , is given by [ theta ] = . for completeness , we also give masses for the two neutral components . their mass degeneracy is lifted by terms in the @xmath46 , [ mass : sp ] m__r^2m__^2+_4+_5v^2/2,m _ _ i^2m__^2+_4-_5v^2/2 . for future convenience , in fig . [ mixing ] we show the mass ratio @xmath47 and @xmath48 for the cases with a very large and normal @xmath49 , respectively . ( solid lines ) and @xmath48 ( solid lines ) on the @xmath50 plane : left @xmath51 tev ; right @xmath52 tev . , title="fig:",width=288 ] ( solid lines ) and @xmath48 ( solid lines ) on the @xmath50 plane : left @xmath51 tev ; right @xmath52 tev . , title="fig:",width=288 ] it is useful to expand the lagrangian eq . ( [ lag ] ) in components . for a more general setup , we introduce a lagrangian that contains a couple of scalar fields @xmath53 with unit charge and as well several majorana fermion @xmath54 instead of only one ( for example in the ma s model there are three rhns ) . their interactions in the mass basis are given by @xmath55,\end{aligned}\ ] ] with @xmath56 the generation index . it is assumed that @xmath54 s are majorana fermions , but practically this assumption is not necessary for generating lfv higgs decay ( but necessary for generating neutrino masses ) . expressed in terms of the original parameters , the couplings can be written as @xmath57 the two neutral sleptons do not play important roles in the following discussions because they do not couple to the higgs boson with a large massive coupling . in this section we will investigate how to get large lfv higgs decay without conflict with the strong constraints such as clfv or @xmath60 . we will concentrate on @xmath58 as an example , but the discussions can be applied to other similar processes . the charged sleptons @xmath62 mediate radiative higgs lfv decay @xmath61 , with the feymann diagram shown in the first panel of fig . [ radiative ] . the corresponding amplitude is generically written as @xmath63 with the form factor @xmath64,\end{aligned}\ ] ] where @xmath65 ; hereafter , we will consider just one flavor of majorana , the dm candidate , and thus the index @xmath66 will be implied . we have neglected the terms proportional to lepton masses , and further assumed @xmath67 in the last line . @xmath68 can be obtained simply by exchanging @xmath69 . we emphasize that to get eq . ( [ fl ] ) which is not suppressed by small lepton masses we need both left- and right - handed scalar mediators , which can be seen obviously from the fact it is proportional to @xmath49-parameter ( see eq . ( [ lag ] ) and also the first panel of fig . [ radiative ] ) . the term with @xmath70 comes from the contributions of @xmath71 and @xmath72 , while the term with @xmath73 comes from those of @xmath74 contributions in the loop . if we had a mediator with only one chirality , the chirality flip required in eq . ( [ eq : h2mt ] ) would occur only in external lepton lines . as a consequence the amplitude would be suppressed by small lepton masses and we could not get sizable @xmath75 rate . in this paper , we follow the notations of three - point scalar function @xmath76 as in ref . @xcite . the loop functions @xmath77 and @xmath78 are defined in eq . ( [ g:12 ] ) and eq . ( [ g:11 ] ) , respectively . as expected , in the decoupling limit with @xmath79 ( or @xmath80 ) , the first term of @xmath81 is suppressed . in contrast , in the maximal mixing limit @xmath82 , the second term is suppressed . later , the former feature will be utilized to suppress lfv decay of charged leptons . the decay width of @xmath83 is calculated as @xmath84 for concreteness , we take @xmath85 and @xmath86 hereafter . in addition , for simplicity we consider only one chiral structure , i.e. , setting @xmath87 . it is easy to recover the corresponding contributions by the replacement @xmath88 and @xmath89 for all the later expressions . the implication of relaxation of this assumption will be commented when necessary . for reference , the branching ratio of @xmath90 is estimated in those two limits , the decoupling limit ( @xmath79 ) : @xmath91 and the maximal mixing limit ( @xmath82 ) : @xmath92 the total decay width of higgs boson has been taken to be 4 mev . we show contour plots of @xmath93 and @xmath94 in fig . [ gx12 ] . ( solid lines ) and @xmath95 ( dashed lines ) . in the plot we use the variables @xmath96 instead of @xmath97 ; the same convention applies to other figures . , width=297 ] the lfv decays of charged leptons are good probes to lfv . for example , the present experimental upper bound on br@xmath99 is @xmath100 @xcite at 90@xmath5 c.l . and will be improved by one order of magnitude in the near future @xcite . the upper bound on clfv decay of muon is even more stringent , br@xmath101 at 90@xmath5 c.l . , from the current meg result @xcite . on the other hand , lfv higgs decay is likely to induce clfv decay ( but not vice versa ) . illustratively , the feynman diagrams of the latter can be obtained simply by replacing the higgs field with a photon leg in the charged loop of the diagram for the former . as a schematic example , see the first and second panels of fig . [ radiative ] . since both processes share almost the same loops , a hierarchical ratio br@xmath102br(@xmath18 ) as large as @xmath103 then raises doubt . lfv decay of @xmath11 into @xmath104 can be generically described by the following effective hamiltonian : @xmath105 different to significant chirality flip by virtue of the higgs field in the loop of lfv higgs decay process , here vector current conserves chirality . there are three other chirality violation sources to generate the wilsonian coefficients @xmath106 , @xmath107.\end{aligned}\ ] ] the expression of @xmath108 can be obtained via @xmath69 . the loop functions @xmath109 and @xmath110 are defined as @xmath111 according to the hamiltonian , the decay width of @xmath112 after summing over polarizations is calculated to be @xmath113.\end{aligned}\ ] ] in @xmath114 , the first and the second terms do not require the simultaneous presence of @xmath115 and @xmath116 because chirality flip comes from the external lines , i.e. , the dirac mass term of lepton . but they require lfv through the same chirality of slepton . these contributions are generically subdominant , compared to the third term , given a large @xmath35 and as well democratic type yukawa coupling , _ i.e. _ , @xmath117 . besides , a sizable mixing angle between @xmath53 and @xmath24 is needed . this means that , not only flavor violation but also chirality violation are provided by the sleptons , as is well understood from the second panel of fig . [ radiative ] . we argue that the @xmath118 rate can be enhanced while suppressing @xmath112 . ( see sec . [ sec : hlfv ] for more details . ) one obvious mechanism is to use heavy @xmath24 , which naturally leads to small mixing angle @xmath119 . in this case the @xmath120 diagram has one more @xmath24 propagator compared with the @xmath121 diagram as shown in the first two diagrams in fig . [ radiative ] , suppressing the former compared to the latter . since lfv higgs decay heavily depends on the charged scalar mixing term , @xmath60 inevitably receives a sizable contribution . under the assumption that other higgs decay modes are not affected , which is a very natural assumption , we get the modification to @xmath60 from the @xmath122loop @xcite loop by assuming much heavier @xmath123 , otherwise the contributions from the @xmath122 and @xmath124loop show substantial cancellation : @xmath125 . then the bound becomes weaker.[degen ] ] , @xmath126 here @xmath127 denotes the reduced coupling of the dimension - five operator for coupling between higgs and photons , @xmath128 . the sign of @xmath49 is indeterminate , so one can make @xmath129 close to the sm value either by requiring a small @xmath130 or @xmath131 , which flips the sign of @xmath127 relative to the sm one . to be more specific , we refer to a recent study @xcite , from which we know that at 68.3@xmath5 c.l . there are two allowed regions : @xmath132feeding these results back to the slepton sector we get the following constraints : * in the first region , one gets the bounds : @xmath133 as one can see , as long as @xmath134 mass is at least a few hundred gevs , the higgs diphoton rate in the decoupling limit can be easily suppressed below the upper bound . but it is not that easy to reconcile br@xmath135 and br@xmath136 in the maximal mixing limit . the @xmath134 should be sufficiently heavy , or it should have roughly equal mass with @xmath123 in the light of footnote [ degen ] . * the second region allows for the scenarios with a huge @xmath49 along with a lighter @xmath137 . in this way of reconciling br@xmath135 and br@xmath136 , it ( asides from determining the sign of @xmath49 ) actually helps to eliminate one of the three parameters in the slepton sector : @xmath138 a tev scale @xmath137 will blow up @xmath49 , thus disfavored . by the way , a too large @xmath139 may also change higgs self - coupling too much . in summary , higgs diphoton does not give a severe constraint . but it is interesting to see that possibly the rate can be related to the large lfv higgs decay . we have collected all the necessary formulas to calculate br@xmath141 under the constraints such as br@xmath142 . in this subsection we show how br@xmath143 can be realized . for that , it is convenient to study the ratio @xmath144 . to explain the central value of the @xmath121 signal , @xmath145 , with the contraint @xmath146 , we need @xmath147 . in the decoupling limit of the scalar system , @xmath148 can be illustratively parameterized as @xmath149 we have made the approximation that eq . ( [ fl ] ) and eq . ( [ cl ] ) are dominated by the second and third terms , respectively . in this approximation , @xmath148 is independent of ( or insensitive to ) the following parameters : ( i ) dm mass @xmath35 ; ( ii ) the yukawa couplings ; ( iii ) to some degree , also @xmath49 . to see the last point , from eq . ( [ theta ] ) one may have @xmath150 and consequently @xmath151 is cancelled . this conclusion holds for a well asymmetric scalar system like @xmath152 , which guarantees decoupling scalars as desired . if instead the scalar sector is in the maximal mixing limit and thus eq . ( [ fl ] ) is dominated by the first term , we have the estimation @xmath153 in the absence of enhancement from ( the inverse of ) small mixing , one needs a huge @xmath49 at least 10 tev and at the same time a very large ratio @xmath154 . while in the previous case it is moderate . that large ratio may incur a significant fine - tuning . in order to lift the ratio , one needs cancelation via introducing some extra contributions to cancel the contribution induced by @xmath155 . in our model this is kind of cancelation happens within well expectation . ] between @xmath156 and @xmath157 . obviously , if @xmath158 , cancelation happens . in particular , for a ( at least ) mild mass hierarchy between @xmath159 and @xmath160 , cancelation approximately determines @xmath35 : @xmath161 ( blue lines ) , which is @xmath162 in the decoupling limit ( left ) and @xmath163 in the maximal mixing limit ( right panel ) . regions with fine - tuning better than @xmath164 are shaded . besides , we label three selected ratios of the masses of two charged scalars ( dashed lines).,title="fig:",width=288 ] ( blue lines ) , which is @xmath162 in the decoupling limit ( left ) and @xmath163 in the maximal mixing limit ( right panel ) . regions with fine - tuning better than @xmath164 are shaded . besides , we label three selected ratios of the masses of two charged scalars ( dashed lines).,title="fig:",width=288 ] regarding @xmath165 as a function of three fundamental variables @xmath166 , we can measure fine - tuning using the quantity @xmath167 explicitly , @xmath168 with @xmath169 , @xmath170 and @xmath171 . let us denote the ratio of loop functions in eq . ( [ rtau ] ) and eq . ( [ rtau : mix ] ) as @xmath172 . in fig . [ ft : r ] , we plot the distributions of @xmath173 and fine - tuning @xmath174 on the @xmath175 plane . the left and the right panel are for the decoupling and the maximal mixing scenarios , respectively . the shaded regions have degree of fine - tuning less than @xmath164 , which is referred as the lower bound for naturalness in this article . it is seen that the decoupling scenario can provide @xmath176 barely incurring fine - tuning ; in contrast , the maximal mixing scenario , which needs @xmath177 , typically incurs fine - tuning worse than @xmath164 . but the cancelation via degenerate @xmath134 and @xmath123 still opens a narrow region around the point @xmath178 or closely alone the line @xmath179 , which without a particular uv reason is not of much interest . in what follows we will focus on these two kinds of natural regions . let us consider the decoupling scenario . we make several observations that are helpful to trace back to the patterns of scalar mass squared matrix . 1 . if both @xmath180 , we need significant degeneracy between two scalars , see the left - bottom corner of the left panel of fig . [ ft : r ] . since we are chasing the decoupling limit confronting a large slepton mixing term with @xmath181 tev , this means large and degenerate scalar mass terms @xmath182 . it results in a heavy spectrum typically having multi - tev sleptons , see the left panel of fig . [ mixing ] . 2 . there is a hierarchy @xmath183 , keeping @xmath184 close to 1 . it requires an asymmetric scalar system , e.g. , @xmath185 , the most favored pattern to decouple @xmath23 and @xmath24 with a large mixing term . @xmath186 , both not far from 1 . this is in the bulk space without special requirements . even for a smaller @xmath49 near the tev scale , one is still able to produce such a case readily , yielding a lighter spectrum inducing dm . in summary , there is a wide parameter space for the decoupling scenario . in practice , in some situations the mixing angle is supposed to be moderately small rather than very small , the dm candidate @xmath19 or the majorana fermion @xmath19 . but only the fermonic dm could be a natural leptophilic dm . ] is a singlet majorana fermion with @xmath15channel mediators , and its phenomenologies in some simplified cases have been investigated compressively in ref . but our case turns out to be significantly different , due to the appearance of both @xmath23 and @xmath24 mediators . in this section we will focus on two main differences , annihilation and direct detection of dm . the first difference comes from dm annihilating . the majorana dm @xmath19 annihilates into leptons through the interactions given in eq . ( [ model : com ] ) . they proceed with @xmath62 exchanging in the @xmath15 and @xmath188channel . we can calculate the cross section expanded in terms of dm relative velocity @xmath189 in the center - of - mass ( cm ) frame : @xmath190 with the @xmath14 and @xmath187wave coefficients respectively given by @xmath191.\end{aligned}\ ] ] the inclusive annihilation rate should sum over the family index @xmath192 and @xmath193 . as a check , when the model goes to the chiral limit considered previously @xcite , e.g. , @xmath194 , we recover the well known result : @xmath195 ( up to contributions suppressed by lepton masses ) . then , dm must annihilate away mainly via @xmath187wave , whose coefficient takes a form of @xmath196 it is not suppressed by small mixing . for instance @xmath197 , it still receives a contribution from @xmath198 . with them , the relic density can be calculated via the well - known formula @xcite @xmath199 at the freeze - out epoch @xmath200 , the effective degree of freedom @xmath201 . if we demand the yukawa coupling constants @xmath202 , in order to maintain perturbativity of the model up to a very high scale , then both dm and mediators should around the weak scale . this is a strong requirement and yields deep implication to direct detection . but here the @xmath14wave may be sufficient to reduce the dm number density , even facing the stringent clfv constraint and at the same time satisfying the tentative lfv higgs decay . it is seen that the @xmath14wave coefficient is directly correlated with clfv decay width @xmath203 , see eq . ( [ cl ] ) . to be more specific , we write @xmath192 in terms of others @xmath204 it may reach the typical cross section of thermal dm , 1 pb . to see this , we parameterize the order of magnitude of @xmath192 as the following : @xmath205 we have taken @xmath206 . therefore , again a weak scale dm along with ( at least one ) weak scale mediator can lead to correct relic density via @xmath14wave annihilation as long as the mixing angle is not highly suppressed . although the @xmath14wave annihilation readily works for flavors like @xmath207 which does not violate lepton flavor , it fails for the case under consideration @xmath208 or inverse . let us show it in the decoupling scenario . with the aid of eq . ( [ br : h1 ] ) and eq . ( [ rtau ] ) we can express @xmath192 as ( aside from the propagator factor ) @xmath209 but that small value of @xmath210 either incurs large fine - tuning or should follow closely the line @xmath211 . the latter leads to additional suppression @xmath212 ( it has been fixed to be 0.15 in the above estimation ) . similarly , the maximal mixing scenario fails either . we make a comment on the coannihilation effect @xcite . despite of not a focus here , it has two interesting points . first , mass degeneracy between @xmath134 and @xmath35 is well consistent with the suppression of br@xmath213 , which is made small by the cancellation mechanism with @xmath214 . for a strong mass hierarchy case @xmath215 , from footnote 4 we have @xmath216 . second , by virtue of a large @xmath40term , the effective cross section of coannihilation is enhanced by the process @xmath217 with @xmath134 in the @xmath15channel : @xmath218 we have worked in the maximal mixing @xmath219 . so , once @xmath220 , the enhancing factor still scales as @xmath221 even for a tev scale @xmath137 . we have shown that dm can gain correct relic density readily . and dm mass should be around the weak scale so as to avoid large yukawa couplings . in this subsection we move to the second difference , direct detection . as a leptophilic dm , dm - nucleon scattering is absent at tree level , but could be generated by radiative corrections . there are two types of corrections leading to dm - nucleon scattering , one mediated by photon and the other higgs boson , respectively . in particular , the second type , which is absent in the previous setup , benefits from @xmath40enhancement and can potentially overcome the loop suppression . the second type is the usual dimension - four operator which comes from the vertex correction on @xmath222 , absent at tree level but generated after ewsb . in the dm direct detection , typically the transferring momentum @xmath223 is very small compared with the other mass scales in the charged particles in the loops , so that @xmath224 where @xmath225 is treated off - shell with invariant mass @xmath226 . the effective coupling at zero momentum transfer is expressed as @xmath227.\end{aligned}\ ] ] using the kinematics and the approximations of two- and three - point scalar functions in appendix [ approx ] , we can further simplify it into @xmath228,\end{aligned}\ ] ] with @xmath229 seen in eq . ( [ g12 ] ) . note that @xmath230 shows cancellation and thus larger @xmath231 dwells on the region with @xmath232 at least modularity larger than @xmath233 . ( solid lines ) in the maximal scenario , setting @xmath51 tev , @xmath234 gev and either @xmath235 or @xmath236 ; again , the variables in this plot are @xmath96 not @xmath97.,width=288 ] the higgs mediated dm - nucleon scattering has a spin - independent cross section @xmath237 with @xmath238 given by @xmath239.\end{aligned}\ ] ] in this paper we take @xmath240 , @xmath241 , @xmath242 and @xmath243 @xcite ( for more discussions about the calculation and uncertainties of these values , see refs . @xcite ) and we then get the estimation @xmath244 , a value near the sensitivity of the current lux . in the bulk parameter space , @xmath245 : @xmath246 the decoupling scenario is hard to be probed , but the maximal mixing scenario , which badly needs a very large @xmath49 , has a good prospect . we choose a benchmark case which is directly related with @xmath58 . photon - mediated scattering becomes important for lighter mediators . since our dm is a majorana fermion , the leading order operator for dm - nucleon coupling is the dimension - six anapole operator @xcite : @xmath247 the @xmath248 can be obtained by integrating out loopy particles step by step @xcite or via direct calculation of the loops @xcite : @xmath249 with @xmath250 . the expression is valid for the heavy leptons with @xmath251 . it is seen that @xmath248 is insensitive to the @xmath40term and the mixing angle . for @xmath252 gev , it is estimated that @xmath253 . the resulting scattering rate is at least four orders of magnitude weaker than the current lux sensitivity @xcite . in sm , lepton flavor is accidentally conserved but on the other hand lfv is an established fact . so it is of importance to search for lfv processes such as lfv higgs decay in the lhc era . it is a good probe to new physics . but lfv higgs decay is negligible and undetectable in most new physics models for addressing neutrino masses . in this paper we study a type of new physics that could lead to large higgs lfv decay , i.e. , lepton - flavored dark matter specified by the particle property of dm ( a majorana fermion ) and dm - sm mediators ( scalar leptons ) . different than other similar setups , here we introduce both the left - handed and the right - handed scalar leptons . they allow for large lfv in higgs decay and thus may explain the tentative br@xmath0 . in particular , we find that the stringent bound from @xmath1 can be naturally avoided especially in the decoupling limit of slepton sector . aspects of relic density and radiative direct detection of the leptonic dm are also investigated . there are several open questions that deserve future investigation . first , as mentioned in the text , neutrino masses and mixings can be radiatively generated because all the core of the ernest ma s model is already incorporated in our model . even restricted to one rhn , i.e. , the majorana dm , we are able to generate realistic neutrino mixings after introducing a couple of scalar lepton doublets @xmath254 . second , in this article we merely discuss lfv in the first and second family of leptons , and such kind of discussions are easily generalized to other families , which is of particular interest when correlated with neutrino phenomenologies . however , it is not easy to reconcile the tiny neutrino mass scale with a large lfv higgs decay like br@xmath0 , because the former basically requires somewhat smaller yukawa couplings @xmath255 . of course , if we work on very light dm like below the gev even mev scale , maybe there still stands a chance . we thank pyungwon ko very much for valuable discussions and reading the manuscript carefully . this work is supported in part by national research foundation of korea ( nrf ) research grant nrf-2015r1a2a1a05001869 ( sb ) . in this appendix we present the technical details used in this paper . the scalar three point function is defined as @xcite . @xmath256 with @xmath257 when @xmath258 , obviously we have @xmath259 . if the invariant masses of the external momentums @xmath260 are far lighter than the mass scales of the particles in the loop , @xmath261 , one can approximate @xmath262 to be @xmath263 with @xmath264 . note that @xmath265 is symmetric under interchanging @xmath266 and @xmath267 . there are two particular limits that are helpful in analyzing the radiative decays of higgs boson . @xmath268 : : for this single propagator case one has @xmath269 if further @xmath266 goes to 1 , it slides to @xmath270 . but for very heavy @xmath271 it decouples as @xmath272 . @xmath273 : : for the asymmetric propagators like this , we have the simple approximation @xmath274 due to the logarithmic factor , it decouples slower than the previous case . and solid lines for @xmath275 ; @xmath276,title="fig:",width=288 ] and solid lines for @xmath275 ; @xmath276,title="fig:",width=288 ] the scalar two - point function is defined as @xmath277 which satisfies the relations @xmath278 and @xmath279 for @xmath258 . actually , it has an explicit expression ( up to @xmath280 ) @xmath281 where @xmath282 and @xmath283 are determined by @xmath284 it has two limits of interest in this paper . let us consider the first limit , i.e. , small external momentum @xmath285 , then we have @xmath286 up to irrelevant additive constants that will be cancelled in the expressions . here @xmath287 with @xmath35 a referred scale . if @xmath288 , one can greatly simplify it into @xmath289 now we move to the other limit , i.e. , when one particle in the loop is extraordinarily lighter than other mass scales ; without loss of generality , let @xmath290 . then one has ( @xmath291 ) @xmath292 in this appendix we derive the approximations of higgs - dm - dm vertex relevant to dm direct detection . the amplitude is given by @xmath293 with @xmath294 where terms suppressed by lepton masses are neglected . for short , we denote @xmath295 . similar conventions are adopted throughout this paper . it does not cause confusion since we have specified an unique index type for each flavor . the vectorial three - point function can be decomposed into @xmath296 after using the motion of equation , one has @xmath297 and @xmath298 . then , the amplitude takes the form of @xmath299 with @xmath300.\end{aligned}\ ] ] @xmath301 is obtained by exchanging @xmath302 and @xmath303 in @xmath304 . specific to the kinematics in this paper , i.e. , @xmath305 , and using the equations below eq . ( [ b0 ] ) and eq . ( [ c0:1 ] ) one can explicitly show @xmath306 . after some exercise one finds the crossed diagram gives @xmath307 . therefore , eventually the form factor relevant to direct detection is @xmath308 . in the @xmath197 limit , the leading order is @xmath227.\end{aligned}\ ] ] note that both the quartic and logarithmic divergencies contained in the two - point functions are cancelled . this is consistent with expectancy and provides as a check for our calculations . it is convenient to write @xmath309 with @xmath310 with @xmath311 . it , again , is in the approximation @xmath312 and @xmath313;. it has an explicit but not illustrative expression , thus not given here . d. aristizabal sierra and a. vicente , phys . d 90 , no . 11 , 115004 ( 2014 ) . j. heeck , m. holthausen , w. rodejohann and y. shimizu , nucl . b * 896 * , 281 ( 2015 ) . y. n. mao and s. h. zhu , arxiv:1505.07668 [ hep - ph ] . i. dorsner , s. fajfer , a. greljo , j. f. kamenik , n. kosnik and i. nisandzic , arxiv:1502.07784 [ hep - ph ] . a. crivellin , g. dambrosio and j. heeck , phys . lett . * 114 * , no . 15 , 151801 ( 2015 ) . f. j. botella , g. c. branco , m. nebot and m. n. rebelo , arxiv:1508.05101 [ hep - ph ] . p. agrawal , z. chacko , c. kilic and c. b. verhaaren , arxiv:1503.03057 [ hep - ph ] . a. hamze , c. kilic , j. koeller , c. trendafilova and j. h. yu , phys . rev . d * 91 * , no . 3 , 035009 ( 2015 ) [ arxiv:1410.3030 [ hep - ph ] ]	in the standard model ( sm ) , lepton flavor violating ( lfv ) higgs decay is absent at renormalizable level and thus it is a good probe to new physics . in this article we study a type of new physics that could lead to large lfv higgs decay , i.e. , a lepton - flavored dark matter ( dm ) model which is specified by a majorana dm and scalar lepton mediators . different from other similar models with similar setup , we introduce both left - handed and right - handed scalar leptons . they allow large lfv higgs decay and thus may explain the tentative br@xmath0 experimental results from the lhc . in particular , we find that the stringent bound from @xmath1 can be naturally evaded . one reason , among others , is a large chirality violation in the mediator sector . aspects of relic density and especially radiative direct detection of the leptonic dm are also investigated , stressing the difference from previous lepton - flavored dm models .
the following result was proved by j.w.s . cassels in 1951 ( see appendix 1 of @xcite ) . [ t1.1]let @xmath0 @xmath1 be sequences of positive real numbers and @xmath2 a sequence of nonnegative real numbers . suppose that@xmath3then one has the inequality@xmath4the equality holds in ( [ 1.2 ] ) when @xmath5 @xmath6 @xmath7 @xmath8 and @xmath9 if one assumes that @xmath10 and @xmath11 for each @xmath12 then by ( [ 2.2 ] ) we may obtain greub - reinboldt s inequality @xcite@xmath13the following unweighted cassels inequality also holds@xmath14provided @xmath15 and @xmath16 satisfy ( 1.1 ) . this inequality will produce the well known plya - szeg inequality @xcite , @xcite:@xmath17provided @xmath10 and @xmath18 for each @xmath19 in @xcite , c.p . niculescu proved , amongst others , the following generalisation of cassels inequality : [ t1.2]let @xmath20 be a vector space endowed with a hermitian product @xmath21 then@xmath22for every @xmath23 and every @xmath24 for which @xmath25 for other reverses of the cauchy - bunyakovsky - schwarz , see the references @xcite-@xcite . in this paper we obtain a generalisation of ( [ 1.6 ] ) for complex numbers @xmath26 and @xmath27 for which @xmath28 applications for isotonic linear functionals , integrals and sequences are also given . the following reverse of schwarz s inequality in inner product spaces holds . [ t2.1]let @xmath29 @xmath30 so that @xmath31 if @xmath32 are such that@xmath33then one has the inequality@xmath34 } { \left [ \func{re}\left ( \overline{a}% a\right ) \right ] ^{\frac{1}{2}}}\leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}% a\right ) \right ] ^{\frac{1}{2}}}\left\vert \left\langle x , y\right\rangle \right\vert . \label{2.2}\]]the constant @xmath35 is sharp in both inequalities . we have , obviously , that@xmath36 -\left\vert x\right\vert ^{2}-\left [ \func{re}\left ( \overline{a}% a\right ) \right ] \left\vert y\right\vert ^{2}\]]and , thus , by ( [ 2.1 ] ) , one has@xmath37 \cdot \left\vert y\right\vert ^{2}\leq \func{re}\left [ a\overline{% \left\langle x , y\right\rangle } + \overline{a}\left\langle x , y\right\rangle % \right ] , \]]giving@xmath38 ^{\frac{1}{2}}}% \left\vert x\right\vert ^{2}+\left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2}}\left\vert y\right\vert ^{2}\leq \frac{\func{re}\left [ a\overline{\left\langle x , y\right\rangle } + \overline{a}\left\langle x , y\right\rangle \right ] } { \left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2}}}. \label{2.3}\]]on the other hand , by the elementary inequality@xmath39holding for @xmath40 and @xmath41 we deduce@xmath42 ^{\frac{1}{2}}}\left\vert x\right\vert ^{2}+\left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{% \frac{1}{2}}\left\vert y\right\vert ^{2}. \label{2.4}\]]utilizing ( [ 2.3 ] ) and ( [ 2.4 ] ) we deduce the first part of ( [ 2.2 ] ) . the last part is obvious by the fact that for @xmath43 @xmath44 now , assume that the first inequality in ( [ 2.2 ] ) holds with a constant @xmath45 i.e. , @xmath46 } { \left [ \func{re}\left ( \overline{a}a\right ) % \right ] ^{\frac{1}{2 } } } , \label{2.5}\]]where @xmath47 and @xmath48 satisfy ( [ 2.2 ] ) . if we choose @xmath49 @xmath50 then obviously ( [ 2.2 ] ) holds and from ( [ 2.5 ] ) we may obtain@xmath51giving @xmath52 the theorem is completely proved . the following corollary is a natural consequence of the above theorem . [ c2.2]let @xmath53 if @xmath32 are such that@xmath54then one has the inequality@xmath55the constant @xmath35 is sharp in ( [ 2.7 ] ) . [ r2.2.1]the inequality ( [ 2.7 ] ) is equivalent to niculescu s inequality ( [ 1.6 ] ) . the following corollary is also obvious . [ c2.3]with the assumptions of corollary [ c2.2 ] , we have@xmath56and@xmath57 ^{2 } \label{2.9 } \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \func{re}\left\langle x , y\right\rangle \right ] ^{2}\leq \frac{\left ( m - m\right ) ^{2}}{4mm}% \left\vert \left\langle x , y\right\rangle \right\vert ^{2}. \notag\end{aligned}\ ] ] if we subtract @xmath58 from the first inequality in ( [ 2.7 ] ) , we get@xmath59which proves the third inequality in ( [ 2.8 ] ) . the other ones are obvious . now , if we square the first inequality in ( [ 2.7 ] ) and then subtract @xmath60 ^{2},$ ] we get@xmath61 ^{2 } & \leq \left [ \frac{\left ( m+m\right ) ^{2}}{4mm}-1\right ] \left [ \func{re}\left\langle x , y\right\rangle % \right ] ^{2 } \\ & = \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \func{re}\left\langle x , y\right\rangle \right ] ^{2}\end{aligned}\]]which proves the third inequality in ( [ 2.5 ] ) . the other ones are obvious . let @xmath62 be an algebra of real functions defined on @xmath63 and @xmath64 a subclass of @xmath62 satisfying the conditions : 1 . @xmath65 implies @xmath66 2 . @xmath67 @xmath68 implies @xmath69 a functional @xmath70 defined on @xmath71 is an _ isotonic linear functional _ on @xmath71 provided that 1 . @xmath72 for all @xmath73 and @xmath74 2 . @xmath75 that is , @xmath76 for all @xmath77 implies @xmath78 the functional @xmath70 is _ normalised _ on @xmath79 provided that @xmath80 i.e. , @xmath81 for all @xmath77 implies @xmath82 usual examples of isotonic linear functionals are integrals , sums , etc . now , suppose that @xmath83 @xmath84 is given and satisfies the properties that @xmath85 @xmath86 @xmath87 for all @xmath88 for a given isotonic linear functional @xmath89 with @xmath90 define the mapping @xmath91 by@xmath92 this functional satisfies the following properties : 1 . @xmath93 for all @xmath94 2 . @xmath95 for all @xmath96 and @xmath97 ; 3 . @xmath98 for all @xmath99 the following proposition holds . [ p3.1]let @xmath100 be such that @xmath85 @xmath101 @xmath102 if @xmath103 are such that@xmath104then for any isotonic linear functional @xmath89 with @xmath105 we have the inequality@xmath106the constant @xmath107 in ( [ 3.3 ] ) is sharp . we observe that@xmath108 \geq 0.\]]applying corollary [ c2.2 ] for @xmath109 we get@xmath110which is clearly equivalent to ( [ 3.3 ] ) . the following additive versions of ( [ 3.3 ] ) also hold . [ c3.2]with the assumption in proposition [ p3.1 ] , one has@xmath111 ^{\frac{1% } { 2}}-a\left ( hfg\right ) \label{3.4 } \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}a\left ( hfg\right ) \notag\end{aligned}\]]and @xmath112 [ r3.3]the condition ( [ 3.2 ] ) may be replaced with the weaker assumption@xmath113 [ r3.4]with the assumption ( [ 3.2 ] ) or ( [ 3.6 ] ) and if @xmath114 with @xmath115 then one has the inequality@xmath116@xmath117 ^{\frac{1}{2% } } -a\left ( fg\right ) \label{3.8 } \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}a\left ( fg\right ) \notag\end{aligned}\]]and@xmath118 let @xmath119 be a measure space consisting of a set @xmath120 a @xmath121algebra @xmath122 of subsets of @xmath27 and a countably additive and positive measure on @xmath122 with values in @xmath123 denote by @xmath124 the hilbert space of all @xmath125-valued functions @xmath126 defined on @xmath27 that are @xmath127integrable on @xmath120 i.e. , @xmath128 where @xmath129 is a measurable function on @xmath130 the following proposition contains a counterpart of the weighted cauchy - bunyakovsky - schwarz s integral inequality . [ p4.1]let @xmath131 @xmath30 with @xmath132 and @xmath133 if @xmath134 \rho \left ( s\right ) d\mu \left ( s\right ) \geq 0 , \label{4.1}\]]then one has the inequality@xmath135 ^{% \frac{1}{2 } } \label{4.2 } \\ & \leq \frac{1}{2}\cdot \frac{\int_{\omega } \rho \left ( s\right ) \func{re}% \left [ a\overline{f\left ( s\right ) } g\left ( s\right ) + \overline{a}f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \notag \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \int_{\omega } \rho \left ( s\right ) f\left ( s\right ) \overline{g\left ( s\right ) } d\mu \left ( s\right ) \right\vert . \notag\end{aligned}\]]the constant @xmath35 is sharp in ( [ 4.2 ] ) . follows by theorem [ t2.1 ] applied for the inner product @xmath136,@xmath137 [ r4.2]a sufficient condition for ( [ 4.1 ] ) to hold is @xmath138 \geq 0\text { \ for } \mu \text{-a.e . \ } s\in \omega . \label{4.3}\ ] ] in the particular case @xmath139 we have the following result . [ c4.3]let @xmath29 @xmath30 with @xmath132 and @xmath140 if@xmath141 d\mu \left ( s\right ) \geq 0 , \label{4.4}\]]then one has the inequality@xmath142 ^{\frac{1}{2 } } \label{4.5 } \\ & \leq \frac{1}{2}\cdot \frac{\int_{\omega } \func{re}\left [ a\overline{% f\left ( s\right ) } g\left ( s\right ) + \overline{a}f\left ( s\right ) \overline{% g\left ( s\right ) } \right ] d\mu \left ( s\right ) } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \notag \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \int_{\omega } f\left ( s\right ) \overline{g\left ( s\right ) } % d\mu \left ( s\right ) \right\vert . \notag\end{aligned}\ ] ] [ r4.4]if @xmath143 then a sufficient condition for either ( [ 4.1 ] ) or ( [ 4.4 ] ) to hold is @xmath144where , in this case , @xmath145 with @xmath146 when @xmath147 are real positive constants , then the following proposition holds . [ p4.5]let @xmath53 if @xmath148 such that@xmath149 d\mu \left ( s\right ) \geq 0 \label{4.7}\]]then one has the inequality@xmath150 ^{% \frac{1}{2 } } \label{4.8 } \\ \leq \frac{1}{2}\cdot \frac{m+m}{\sqrt{mm}}\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) .\end{gathered}\ ] ] the proof follows by corollary [ c2.2 ] applied for the inner product @xmath151 the following additive versions also hold . [ c4.6]with the assumptions in proposition [ p4.5 ] , one has the inequalities@xmath152 ^{\frac{1}{2 } } \label{4.9 } \\ & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } % \right ] d\mu \left ( s\right ) \notag \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}\int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \notag\end{aligned}\]]and @xmath153 d\mu \left ( s\right ) \right ) ^{2 } \notag \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left ( \int_{\omega } \rho \left ( s\right ) \func{re}\left [ f\left ( s\right ) \overline{g\left ( s\right ) } \right ] d\mu \left ( s\right ) \right ) ^{2}. \notag\end{aligned}\ ] ] [ r4.7]if @xmath154 , a sufficient condition for ( [ 4.1 ] ) to hold is @xmath155where @xmath156 for a given sequence @xmath157 of nonnegative real numbers , consider the hilbert space @xmath158 @xmath159 where @xmath160 [ p5.1]let @xmath29 with @xmath161 and @xmath162 if@xmath163 \geq 0 , \label{5.2}\]]then one has the inequality@xmath164 ^{\frac{% 1}{2 } } & \leq \frac{1}{2}\cdot \frac{\sum_{i=0}^{\infty } w_{i}\func{re}\left [ a\overline{x_{i}}y_{i}+\overline{a}x_{i}\overline{y_{i}}\right ] } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \label{5.3 } \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \sum_{i=0}^{\infty } w_{i}x_{i}\overline{y_{i}}\right\vert . \notag\end{aligned}\]]the constant @xmath35 is sharp in ( [ 5.3 ] ) . [ c5.3]let @xmath29 with @xmath161 and @xmath168 if@xmath169 \geq 0 , \label{5.4b}\]]then one has the inequality@xmath170 ^{\frac{1}{2}% } & \leq \frac{1}{2}\cdot \frac{\sum_{i=0}^{\infty } \func{re}\left [ a% \overline{x_{i}}y_{i}+\overline{a}x_{i}\overline{y_{i}}\right ] } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1}{2 } } } \label{5.5 } \\ & \leq \frac{1}{2}\cdot \frac{\left\vert a\right\vert + \left\vert a\right\vert } { \left [ \func{re}\left ( \overline{a}a\right ) \right ] ^{\frac{1% } { 2}}}\left\vert \sum_{i=0}^{\infty } x_{i}\overline{y_{i}}\right\vert . \notag\end{aligned}\ ] ] [ p5.5]let @xmath53 if @xmath173 such that @xmath174 \geq 0 , \label{5.7}\]]then one has the inequality@xmath175 ^{\frac{% 1}{2}}\leq \frac{1}{2}\cdot \frac{m+m}{\sqrt{mm}}\sum_{i=0}^{\infty } w_{i}% \func{re}\left ( x_{i}\overline{y_{i}}\right ) . \label{5.8}\ ] ] [ c5.6]with the assumptions in proposition [ p5.5 ] , one has the inequalities@xmath177 ^{\frac{% 1}{2}}-\sum_{i=0}^{\infty } w_{i}\func{re}\left ( x_{i}\overline{y_{i}}\right ) \label{5.9 } \\ & \leq \frac{\left ( \sqrt{m}-\sqrt{m}\right ) ^{2}}{2\sqrt{mm}}% \sum_{i=0}^{\infty } w_{i}\func{re}\left ( x_{i}\overline{y_{i}}\right ) \notag\end{aligned}\]]and@xmath178 ^{2 } \label{5.10 } \\ & \leq \frac{\left ( m - m\right ) ^{2}}{4mm}\left [ \sum_{i=0}^{\infty } w_{i}% \func{re}\left ( x_{i}\overline{y_{i}}\right ) \right ] ^{2}. \notag\end{aligned}\ ] ]	a generalisation of the cassels and greub - reinboldt inequalities in complex or real inner product spaces and applications for isotonic linear functionals , integrals and sequences are provided .
the variational and diffusion quantum monte carlo methods ( vmc and dmc ) are zero - temperature stochastic techniques for evaluating the expectation values of time - independent operators @xcite . these methods are particularly well suited for calculating the ground - state energies of large assemblies of interacting quantum particles . the central object is an approximate trial wave function whose accuracy controls the computed energy and the intrinsic statistical fluctuations in the calculations . it is therefore of particular importance to develop accurate trial wave functions . expectation values in vmc are evaluated using importance - sampled monte carlo integration . in dmc the ground state is projected out by evolving the schrdinger equation in imaginary time . such projector methods suffer from a fermion sign problem , in which errors in the propagation increase exponentially in imaginary time as the algorithm amplifies any spurious component of the lower - energy bosonic state . this problem may be evaded in dmc by employing the fixed - node approximation @xcite , in which the nodal surface is fixed to be that of a suitable trial wave function . the resulting dmc energy is greater than or equal to the exact energy and less than or equal to the vmc energy computed with the same trial wave function . the vmc energy depends on the entire trial wave function , but the dmc energy depends only on the nodal surface of the trial wave function . one of the appealing features of vmc and dmc is that virtually any form of trial wave function can be used . the main criteria are that the wave function must obey the correct symmetry under particle exchange , it should be flexible enough to describe the system of interest , and that it should be possible to evaluate it rapidly . the analytic properties and normalizability of the trial wave function must be such that the energy expectation value is well - defined . the simplest fermionic wave function is a slater determinant , which describes exchange but not correlation . multideterminant wave functions , pairing wave functions such as geminals @xcite , and backflow transformations @xcite can also be used . the most fruitful method of going beyond the slater determinant is , however , to multiply it by a jastrow factor @xcite , which leads to the slater - jastrow wave function . the jastrow factor is normally chosen to depend on the interparticle separations , which introduces correlation into the wave function . the introduction of a jastrow factor often leads to the recovery of 80% or more of the correlation energy of electronic systems @xcite . the jastrow factor is chosen to be everywhere positive and symmetric with respect to the exchange of identical particles in order to maintain the nodal surface defined by the rest of the wave function . one of the features of the jastrow factor is that it can conveniently be used to enforce the kato cusp conditions @xcite , which determine the behavior of the wave function when two charged particles approach one another . enforcing the kato cusp conditions does not necessarily improve the variational energy , but the reduction in the statistical fluctuations in the energy is often very important . dmc can be viewed as vmc with a perfect jastrow factor , but improving the jastrow factor can improve dmc calculations in several ways . the dmc algorithm is subject to time - step errors and to ( normally very small ) population - control errors @xcite that are reduced by improving the trial wave function . evaluating expectation values of operators that do not commute with the hamiltonian is not straightforward in dmc , but using highly accurate trial wave functions helps in achieving more accurate results . similar considerations apply when using nonlocal pseudopotentials , which involves making approximations that are ameliorated by improving the trial wave function @xcite . as the fundamental limitation on the accuracy of dmc is the quality of the nodal surface , it is desirable to use trial wave functions with optimizable nodal surfaces as afforded by , for example , multideterminant wave functions and backflow transformations . a good jastrow factor can account for the bulk of the dynamical correlation energy , which allows the optimization of parameters that affect the nodal surface to achieve a better nodal surface . here we introduce a highly flexible form of jastrow factor which allows for the introduction of a variety of terms involving arbitrary numbers of particles . our main motivation is to be able to implement quickly different functional forms and explore the importance of different correlations in any physical system we study . jastrow factors correlating several electrons have been used in earlier calculations , such as those of refs . we study the effects of various three - body jastrow terms and introduce a four - body van der waals - like term . we also construct anisotropic jastrow factors that can capture the natural symmetries of a system . we have successfully applied these jastrow factors to a variety of systems , and we report results for the one- and two - dimensional homogeneous electron gases , the be , b , and o atoms , and the beh , n@xmath0 , h@xmath0o , and h@xmath0 molecules . our vmc and dmc calculations were performed using the casino package @xcite . hartree atomic units are used throughout ( @xmath1 ) . the structure of the paper is as follows . we describe the form and properties of the general jastrow factor in sec . [ sec : gjastrow ] . specific examples of the construction of jastrow terms are given in sec . [ sec : construction ] , and results obtained using them are presented in sec . [ sec : results ] . finally , we draw our conclusions in sec . [ sec : conclusions ] . implementation details are given in appendix [ app : implementation ] , and an example can be found in appendix [ app : example ] . we report only summaries of our data in this paper @xcite . quantum monte carlo ( qmc ) methods can be applied to systems which can be generically described as an ensemble of @xmath2 quantum particles and @xmath3 sources of external potential . the most common type of qmc simulations are electronic calculations , where the quantum particles are electrons and the sources of external potential are fixed nuclei ( or pseudopotentials ) . for simplicity , we refer to quantum particles as electrons and to external potentials as nuclei in the rest of this paper . our jastrow factor is applicable to other types of quantum particles and external potentials . any trial wave function can be written in the form @xmath4 where @xmath5 is the part of the wave function that imposes the symmetry and boundary conditions , and @xmath6 is a jastrow correlation factor which is constrained so that the symmetry and boundary properties of @xmath5 are transferred unmodified to @xmath7 . in the single - determinant ( sd ) slater - jastrow wave function , @xmath5 is a slater determinant . there are various alternatives to slater determinants , with different advantages and disadvantages depending on the system . typically @xmath8 is constructed as a sum of terms , e.g. , @xmath9 where `` e e '' stands for `` electron - electron , '' `` e n '' for `` electron nucleus , '' etc . each of these terms involves different numbers of electrons @xmath10 and nuclei @xmath11 . we shall refer to @xmath10 and @xmath11 as the electronic and nuclear ranks of a term , respectively , which are constrained to satisfy @xmath12 , @xmath13 , and @xmath14 . we have designed a generic jastrow term of selectable ranks , @xmath15 , such that the total jastrow factor is constructed as the exponential of a sum of one or more terms of the desired ranks . in this notation , @xmath16 , @xmath17 , etc . the function @xmath15 is a sum over all sets of @xmath10 electrons and @xmath11 nuclei in the system of a parameterized function of the e e and e n relative position vectors within each such set . while alternatives exist , a natural way of parameterizing this function for arbitrary values of @xmath10 and @xmath11 ( implying an arbitrary number of variables in the function ) is to expand it in products of functions of the individual e e and e n vectors . thus , we construct our jastrow factor using pairwise objects as building blocks , and in what follows we describe these objects and derive the properties of @xmath15 that follow from those of the pairwise objects . we name the e e functions used in the expansion @xmath18 , where @xmath19 is the relevant e e relative position vector , @xmath20 is the index of the function within a chosen basis of functions , and @xmath21 is the e e dependency index , which allows the use of different optimizable parameters , if present , for parallel- and antiparallel - spin electron pairs , for example . similarly , the e n basis functions are @xmath22 , where @xmath19 is the relevant e n relative position vector , @xmath23 is the index of the function within the chosen basis set , and @xmath24 is the e n dependency index of the basis set , which allows the use of different parameters for up- and down - spin electrons around a given nucleus , or for different atoms , for example . in the case of nonelectronic systems , e e and e n dependency indices are used to distinguish between particle types and spins . we introduce a compact notation for defining @xmath15 . we represent the @xmath10 electronic indices by the integer vector @xmath25 , each of whose components takes a distinct value between @xmath26 and @xmath2 , and the @xmath11 nuclear indices by the integer vector @xmath27 , each of whose components takes a distinct value between @xmath26 and @xmath3 . for each term in the jastrow factor we define the e e and e n dependency matrices @xmath28 and @xmath29 of respective sizes @xmath30 and @xmath31 containing the dependency indices @xmath32 and @xmath33 for each e e and e the components of @xmath28 and @xmath29 can be made equal depending on the symmetries of the system , including particle distinguishability and geometrical symmetries which make different nuclei equivalent . likewise , it is convenient to use matrices to represent the basis functions involved in the jastrow factor term . for e e basis functions , each row and column of the @xmath34 matrix @xmath35 corresponds to an electron , @xmath36 we refer to the @xmath37 matrix formed by the e e dependency indices @xmath38 as @xmath39 . both @xmath39 and the @xmath34 matrix of e e expansion indices @xmath40 are defined to be symmetric , and this fact has been used in eq . ( [ eq : def_phimat ] ) . noting that @xmath41 , and restricting the e e functions to be either symmetric or antisymmetric about the origin , one finds in eq . ( [ eq : def_phimat ] ) that matrix @xmath35 is symmetric , antisymmetric , or asymmetric depending on whether the functions in the basis set are all symmetric , all antisymmetric , or both types are present , respectively . for e n basis functions each row of the @xmath42 matrix @xmath43 corresponds to an electron and each column to a nucleus , @xmath44 we refer to the @xmath45 matrix formed by the e n dependency indices @xmath46 as @xmath47 , and the @xmath45 matrix of e n expansion indices is @xmath48 . we write @xmath49 as a sum of contributions from each group of @xmath10 electrons and @xmath11 nuclei in the system , @xmath50 where summations with vector indices represent sums in which every component of the vector is a summation index , and `` @xmath51 '' ( for `` sorted vector '' ) indicates that the sum is restricted to vectors whose components are sorted , e.g. , @xmath52 , which avoids redundant contributions @xcite . the contribution from the @xmath10-electron and @xmath11-nucleus group @xmath53 is @xmath54 where @xmath55 are the linear parameters , summations with matrix indices represent sums in which every component of the matrix is a summation index , @xmath56 acting on matrices implies the product of all of their components , and `` u.t . '' means that the relevant operation is restricted to the upper - triangular portion of the e e matrices involved , excluding the diagonal . equation ( [ eq : gjastrow_sum_of_core ] ) imposes the condition that @xmath57 must not depend on the specific ordering of the electrons and nuclei listed in @xmath58 and @xmath59 . let @xmath60 and @xmath61 be permutation matrices of respective sizes @xmath34 and @xmath62 such that @xmath63 and @xmath64 are integer vectors containing reordered electronic and nuclear indices . the value of @xmath65 should therefore equal that of @xmath57 , @xmath66 \prod^{\rm u.t . } { { \underline { \phi}}}_{{\underline { \nu}}}^{{\underline { \rm p}}}({{\bf i } } ) \prod { { \underline { \theta}}}_{{\underline { \mu}}}^{{\underline { \rm s}}}({{\bf i}},{{\bf i } } ) \;,\end{aligned}\ ] ] where @xmath67 = \frac{\prod^{\rm u.t.}{{\underline { \rm a}}}{\,}{{\underline { \phi}}}_{{\underline { \nu}}}^{{\underline { \rm p}}}({{\bf i}}){\,}{{\underline { \rm a}}^{\rm t } } } { \prod^{\rm u.t.}{{\underline { \phi}}}_{{\underline { \nu}}}^{{\underline { \rm p}}}({{\bf i } } ) } \;,\ ] ] which is @xmath68 for basis sets consisting only of symmetric functions , while in the presence of antisymmetric basis functions it may be @xmath68 or @xmath69 depending on the precise permutation performed by @xmath60 . equating the right - hand sides of eqs . ( [ eq : gjastrow_core ] ) and ( [ eq : gjastrow_core_permute ] ) one finds that @xmath70 \lambda_{{{\underline { \rm a}}}{\,}{{\underline { \nu}}}{\,}{{\underline { \rm a}}^{\rm t } } , { { \underline { \rm a}}}{\,}{{\underline { \mu}}}{\,}{{\underline { \rm b}}^{\rm t}}}^ { { { \underline { \rm a}}}{\,}{{\underline { \rm p}}}({{\bf i}}){\,}{{\underline { \rm a}}^{\rm t } } , { { \underline { \rm a}}}{\,}{{\underline { \rm s}}}({{\bf i}},{{\bf i}}){\,}{{\underline { \rm b}}^{\rm t } } } \;.\ ] ] this equation represents the basic symmetry property of the linear parameters of the jastrow factor , which implies that a parameter with a given set of superindices @xmath71 is determined by another parameter with a permuted set of superindices @xmath72 . this redundancy is removed by considering only one of the possible permutations of @xmath71 . we call this particular permutation of @xmath73 the signature of the group of particles @xmath53 , @xmath74 where the permutation matrices @xmath75 are computed by applying a matrix - sorting algorithm @xcite to @xmath71 . in our terminology , the set of linear parameters whose superindices reduce to the same signature constitute a parameter channel . only those parameters whose superindices equal the signature of a channel need be stored , and any other linear parameters in the channel can be computed from them via eq . ( [ eq : lambda_symmetry ] ) . the signature @xmath76 may contain repeated entries such that there exist permutation matrices @xmath77 that leave the signature unchanged , @xmath78 in which case eq . ( [ eq : lambda_symmetry ] ) becomes @xmath79 \lambda_{{{\underline { \rm a}}}{\,}{{\underline { \nu}}}{\,}{{\underline { \rm a}}^{\rm t } } , { { \underline { \rm a}}}{\,}{{\underline { \mu}}}{\,}{{\underline { \rm b}}^{\rm t}}}^{{{\underline { \tilde { \rm p}}}}({{\bf i}}),{{\underline { \tilde { \rm s}}}}({{\bf i}},{{\bf i } } ) } \;.\ ] ] equation ( [ eq : lambda_symmetry_invariant_signature ] ) is the symmetry constraint that relates linear parameters within a channel , which can be imposed as detailed in sec . [ sec : constraints ] . the components of @xmath40 and @xmath48 are the e e and e n expansion indices . we define the expansion indices so that they can each take any value between 1 and the e e expansion order @xmath80 , and between 1 and the e n expansion order @xmath81 , respectively . we factorize an optional cutoff function into @xmath82 and @xmath83 , so that @xmath84 for @xmath85 , and @xmath86 for @xmath87 , where @xmath88 and @xmath89 are the e e and e n cutoff functions and @xmath90 and @xmath91 are functions from a suitable basis set . this factorization allows an efficient implementation of localized jastrow factor terms . additionally , we allow expansion indices to take a value of zero with the special meaning that @xmath92 for all @xmath21 , @xmath24 , and @xmath19 . note that these 0-th functions do not contain cutoff functions . this allows us to construct terms with specialized functional forms , such as those involving dot products of vectorial quantities . constraints on the parameters can be expressed in the form of a system of equations involving the linear parameters and the basis function parameters . we restrict our analysis to linear constraints on the linear parameters , and constraints that can be imposed on the nonlinear parameters contained in a basis function independently from the linear parameters and from nonlinear parameters in other basis functions . linear constraints on the linear parameters can be imposed using gaussian elimination , as described in ref . the matrix of coefficients may depend on the nonlinear parameters in the basis functions , if present , and the linear system is usually underdetermined , resulting in a subset of the parameters being determined by the values of the remaining parameters , which can be optimized directly . when a constraint results in setting specific linear parameters to zero , it is more convenient simply to remove them from the list of linear parameters . this is accomplished by disallowing the indices @xmath40 and @xmath48 from taking the values corresponding to linear parameters . we call this an indexing constraint . symmetry constraints must always be imposed , otherwise the trial wave function is unphysical and calculations give erroneous results . symmetry constraints amount to equalities between pairs of parameters as per eq . ( [ eq : lambda_symmetry_invariant_signature ] ) . when two of these equalities relate the same pair of parameters with opposite signs , e.g. , @xmath93 and @xmath94 , which implies @xmath95 , both parameters are eliminated using indexing constraints . the coulomb potential energy diverges when the positions of two electrons or an electron and a nucleus coincide . however , the local energy of an eigenstate of the hamiltonian , including the exact ground - state wave function , is finite and constant throughout configuration space . divergences in the local energy are therefore not a feature of the exact wave function and can lead to poor statistics in qmc calculations ; hence it is important to avoid them . the kinetic energy must diverge to cancel out the potential energy and keep the local energy finite , which is achieved by demanding that the wave function obeys the kato cusp conditions @xcite . for any two charged particles @xmath96 and @xmath97 in a two- or three - dimensional system , these are @xmath98 where @xmath99 denotes the spherical average of @xmath100 , @xmath81 represents charge , @xmath101 is the reduced mass , @xmath11 represents mass , @xmath102 is the dimensionality , and the positive sign in the denominator is for indistinguishable particles and the negative sign is for distinguishable particles . fixed nuclei are regarded as having an infinite mass . divergent interactions other than the coulomb potential would give rise to different expressions . it is common practice to impose the e n cusp conditions on @xmath103 and the e e cusp conditions on the jastrow factor . this is because typical forms of @xmath103 explicitly depend on the e n distances but not on the e e distances . our implementation allows the option of applying both types of cusp conditions to the jastrow factor , which gives flexibility in the choice of @xmath103 and its properties . in particular , we impose the cusp conditions on a single jastrow factor term , and constrain all other terms in the jastrow factor so that their contribution to the local kinetic energy is finite at e e and e n coalescence points . for nondivergent interaction potentials , such as most pseudopotentials , we simply require that the kinetic energy remains finite at coalescence points . our implementation is also capable of not applying any constraints at e e and e n coalescence points since this is advantageous in some cases @xcite . imposing that the kinetic energy be finite at coalescence points is nontrivial if the jastrow factor contains anisotropic functions . consider the exponent of a jastrow factor @xmath104 near a point where two particles coalesce , be it two electrons or an electron and a nucleus . the dependence of @xmath104 on coordinates other than those of the coalescing particles should be smooth in the vicinity of the coalescence point , and therefore one should be able to write @xmath105 , where @xmath19 is the difference between the position vector of the two particles , and all remaining particles are held fixed . the local kinetic energy is computed from two estimators , one involving @xmath106 and the other @xmath107 . we require both quantities to remain finite as @xmath108 . we expand the jastrow factor in spherical harmonics , @xmath109 , with @xmath110 , and the gradient and laplacian of @xmath111 are @xmath112 @xmath113 y_{l , m}(\theta,\phi ) \;.\end{aligned}\ ] ] let us assume that @xmath114 is finite at the origin and expand it to second order about @xmath115 , @xmath116 . substituting into eqs . ( [ eq : def_grad_j ] ) and ( [ eq : def_lap_j ] ) , and ignoring contributions of @xmath117 or higher , we arrive at @xmath118 \frac{\partial y_{l , m}(\theta,\phi)}{\partial \theta } { \bf u}_\theta \nonumber \\ & & + \left [ \frac{f_{l , m}(0)}{r\sin\theta } + \frac{f_{l , m}^\prime(0)}{\sin\theta } \right ] \frac{\partial y_{l , m}(\theta,\phi)}{\partial \phi } { \bf u}_\phi \;,\end{aligned}\ ] ] @xmath119 y_{l , m}(\theta,\phi ) \;.\end{aligned}\ ] ] the coefficient of the negative powers of @xmath120 in eqs . ( [ eq : grad_j ] ) and ( [ eq : lap_j ] ) must vanish for @xmath121 and @xmath122 to be finite at the coalescence point . this gives rise to two conditions : ( a ) if @xmath123 then @xmath124 ; and ( b ) if @xmath125 then @xmath126 . application of the kato cusp or finite kinetic energy constraints requires the construction of a linear system for each linear - parameter channel in each term of the jastrow factor based on the above equations . let @xmath127 be an operator such that @xmath128 . the cusp equations associated with the coalescence of electrons @xmath96 and @xmath97 have the form @xmath129 } { \partial r_{ij } } \right]_{r_{ij}=0 } = \gamma_{ij } \;,\ ] ] where @xmath130 is the right - hand side of eq . ( [ eq : kato ] ) . the label `` @xmath131 '' ( for `` one contribution '' ) denotes that the sum is restricted to values of the @xmath40 sets such that the elements of the upper triangular portion of @xmath132 are all 1 except that corresponding to the electron pair formed by @xmath96 and @xmath97 , and the label `` @xmath133 '' ( for `` no contribution '' ) denotes that the sum is restricted to values of @xmath48 such that all elements of @xmath134 are 1 . these restrictions are trivially satisfied by e e terms . parameters that do not contribute to eq . ( [ eq : ee_cusp ] ) should be set by the condition that the kinetic energy does not diverge at e e coalescence points , resulting in @xmath135 \right]_{r_{ij}=0 } = 0 \;,\ ] ] for @xmath123 , and @xmath136 } { \partial r_{ij } } \right]_{r_{ij}=0 } = 0 \;,\ ] ] for @xmath125 . the anisotropy of @xmath137 at @xmath138 determines which of eqs . ( [ eq : ee_noaniso ] ) and eqs . ( [ eq : ee_nocusp ] ) need be imposed . the label @xmath139 ( for `` equal - product '' ) denotes that the sum is only over indices associated with electrons @xmath96 and/or @xmath97 , and these indices take only values such that the product of the pair of functions associated with @xmath140 and @xmath141 ( @xmath142 and @xmath143 in the e n case ) is equal throughout the sum . each set of distinct two - function products and each value of @xmath144 correspond to different equations , and each set of possible values of the indices not summed over corresponds to a separate set of equations . for the coalescence of electron @xmath96 and nucleus @xmath145 the cusp conditions take the form @xmath146 } { \partial r_{ii } } \right]_{r_{ii}=0 } = \gamma_{ii } \;,\ ] ] while the kinetic energy is kept finite if @xmath147 \right]_{r_{ii}=0 } = 0 \;,\ ] ] for @xmath123 and @xmath148 } { \partial r_{ii } } \right]_{r_{ii}=0 } = 0 \;,\ ] ] for @xmath125 . the equal - product constraint on the sum is now such that the sum is only over e e indices associated with electron @xmath96 and e n indices associated with nucleus @xmath145 , and these indices only take values such that the product of the pair of functions associated with @xmath140 and @xmath149 is equal throughout the sum . again , each set of distinct two - function products corresponds to a different equation , and each set of possible values of the indices not being summed over corresponds to a separate set of equations . note that the equal - product constraints in the sums of eqs . ( [ eq : ee_cusp ] ) and ( [ eq : en_cusp ] ) reduce to the equal - product constraints described for the e n @xmath150 term of the drummond - towler - needs ( dtn ) jastrow factor in the appendix of ref . @xcite when natural - power basis functions are chosen . it is possible to construct terms containing dot products by using appropriate constraints . for example , consider the basis functions @xmath151 , @xmath152 , and @xmath153 . in an e n term we can restrict the indices so that @xmath48 takes only the values @xmath154 , @xmath155 , @xmath156 , so that the contribution of electron @xmath96 and nuclei @xmath145 and @xmath104 is @xmath157 , provided we also apply a linear constraint that equates the three nonzero linear coefficients . section [ subsec : specific_terms ] gives a practical example of a term containing dot products , which is used in sec . [ subsec : h2_mol ] . it is also possible to introduce boys - handy - style indexing , @xcite where the sum of all e e and e n indices is restricted to be less than or equal to some fixed integer @xmath158 . this is accomplished by setting the e e and e n expansion orders to @xmath158 and then eliminating the parameters that violate the conditions via indexing restrictions . possibly the simplest basis set is the natural powers , @xmath159 as used in the dtn jastrow factor for the localized @xmath160 , @xmath161 , and @xmath150 terms @xcite . these functions need to be cut off at some radius @xmath162 , for which purpose the dtn jastrow factor uses the polynomial cutoff function @xmath163 where @xmath162 is an optimizable parameter , @xmath164 is a positive integer , and @xmath165 is the heaviside step function . we also use a slightly different version of this cutoff function , @xmath166 which should be numerically superior to @xmath167 . a particular variant of @xmath168 is the anisotropic cutoff function @xmath169^{p^{(i)}_\beta } \;,\ ] ] where @xmath162 is an optimizable parameter , @xmath164 is a positive integer , @xmath102 is the dimensionality of the system , @xmath170 are unit vectors along @xmath102 orthogonal directions , @xmath171 are real - valued constants , and @xmath172 are integer exponents , which are constrained so that @xmath173 is the same for all values of @xmath96 . this cutoff function is simply the product of an isotropic cutoff function and a spherical harmonic . for example , with @xmath174 , @xmath175 , @xmath176 , and @xmath177 , and the vectors pointing along the cartesian axes , we obtain @xmath178 \;,\ ] ] which is proportional to a real spherical harmonic with @xmath179 . the advantage of describing anisotropy in the cutoff function rather than in the basis functions is that the common spherical harmonic can be factorized out of the sum over expansion indices , which reduces the computational cost . we allow different orientations to be used for different e n dependency indices , which is useful to adapt the functional form to , e.g. , the geometry of a molecule . an alternative to the natural - power basis in finite systems is a basis of powers of fractions which tend to a constant as @xmath180 , and therefore do not need to be cut off . we define the basis @xmath181 where @xmath182 and @xmath183 are real - valued optimizable parameters . similar basis sets with @xmath184 have been used in the literature , often in conjunction with boys - handy - style indexing @xcite , and this basis was used in ref . @xcite with an early implementation of the jastrow factor presented here . in extended systems it is important to use a basis that is consistent with the geometry of the simulation cell and has the periodicity of the system , such as a cosine basis , @xmath185 where the @xmath186 vectors are arranged in stars defined by the cell geometry . this basis is used in the dtn jastrow factor for the extended @xmath80 and @xmath81 terms . a suitable basis set for building specialized terms containing dot products is @xmath187 } \frac { { \bf r } \cdot { \hat{\bf u}_{{\rm mod}\left(\nu-1,d\right)+1 } } } { r } \;,\ ] ] where @xmath102 is the dimensionality of the system and @xmath188 are unit vectors parallel to the @xmath102 cartesian axes . a term constructed using these functions with appropriate index - restriction constraints would consist of dot products between two vectors multiplied by a natural - power expansion in their moduli . we employ a condensed notation to refer to jastrow terms that use certain basis functions , cutoff functions and constraints . each term is represented by a single capital letter , with @xmath10 and @xmath11 as subindices . any other relevant information is given as a superindex . typically we use expansion orders @xmath80 and @xmath81 of 79 for two - body terms , 45 for three - body terms , and 23 for four - body terms , except when indicated otherwise . for simple jastrow terms we use the natural power basis functions @xmath189 and the polynomial cutoff functions @xmath21 or @xmath190 . we refer to these terms as @xmath191 . @xmath192 , @xmath193 , and @xmath194 are the equivalent of the dtn @xmath160 , @xmath161 , and @xmath150 terms , respectively . in the @xmath194 term , and in any term where more than one electron and one or more nuclei are involved , we choose not to apply e e cutoff functions , relying instead on the e n cutoffs to fulfill this role . additional @xmath191 terms used in this paper that were not part of the dtn jastrow factor are @xmath195 , @xmath196 , @xmath197 , @xmath198 , @xmath199 , and @xmath200 . in @xmath191 we typically use a truncation order in the cutoff function of @xmath201 . we use @xmath202 to refer to the anisotropic variant of @xmath191 . the @xmath202 term consists of natural power basis functions @xmath189 and the anisotropic cutoff function @xmath203 , and `` @xmath204 '' is a placeholder for describing the spherical harmonic . for example , for the highly anisotropic n@xmath205 molecule we use terms such as @xmath206 , @xmath207 , @xmath208 , and @xmath209 . in finite systems we also use the @xmath210 basis functions in terms without explicit cutoff functions which we call @xmath211 , or @xmath212 when we force @xmath184 in the basis functions . in some systems it is useful to apply boys - handy - style indexing to @xmath212 , and we refer to the resulting term as @xmath213 . in extended systems we make use of the cosine basis functions @xmath214 in terms denoted @xmath215 , where we choose expansion orders so that at least as many @xmath216 vectors as electrons in each spin channel are included in the expansion . to test the flexibility of our implementation we have designed an e e n n jastrow term for describing the correlations associated with van der waals interactions , which we call @xmath217 . this term is capable of distinguishing between configurations where the electron - nucleus relative position vectors @xmath218 and @xmath219 are parallel from those where they are antiparallel . introducing a dot product achieves this effect , and @xmath217 has the following functional form , @xmath220 we require basis functions to be scalars in our jastrow factor , so the dot product is separated into its components . hence , we construct the @xmath217 term using @xmath221 for the e n basis with @xmath21 as the e n cutoff functions , and @xmath189 for the e e basis . we allow e n indices to be zero , and apply a number of constraints on the linear parameters : ( a ) we eliminate all index sets except those in which the e n indices are of the form @xmath222 or @xmath223 , with @xmath224 ; ( b ) we eliminate all parameters that do not satisfy @xmath225 ; ( c ) we equate any two linear parameters @xmath226 and @xmath227 if the zeros of @xmath228 and @xmath229 are in the same position and their nonzero components satisfy @xmath230 = { \rm int}\left[(k_2 - 1)/3\right]$ ] and @xmath231 = { \rm int}\left[(l_2 - 1)/3\right]$ ] . these constraints are applied in addition to the generic symmetry and cusp constraints . table [ tab : notation ] summarizes the notation for the terms we have introduced . [ cols="<,<,<,<",options="header " , ] ref . @xcite . ref . @xcite . ref . @xcite ( using a slightly different bond length ) . ref . we have described a generalized jastrow factor allowing terms that explicitly correlate the motions of @xmath10 electrons with @xmath11 static nuclei . these terms can be parameterized using various basis sets , including terms that involve dot products of interparticle position vectors . we have also introduced anisotropic cutoff functions . the formalism may be applied to systems with particle types and external potentials other than electrons and coulomb potentials . optimization of the wave function is one of the most human- and computer - time consuming tasks in performing qmc calculations . we have performed term - by - term optimizations to understand how different terms in the jastrow factor contribute to the electronic description of a system , and we hope that our analysis will serve as a guideline for constructing jastrow factors for other systems . we have tested these terms on hegs , atoms , and molecules . the variational freedom from the higher - order terms generally improves the quality of the wave function . we have only considered single - determinant wave functions in this study , although our jastrow factor can of course be used with other wave function forms . we have demonstrated the construction and application of an e e n n jastrow factor term designed to describe van der waals interactions between atoms . this term retrieves a large fraction of the van der waals correlation energy in tests on the triplet state of h@xmath0 at the proton separation that minimizes the total energy of the system . we have found evidence for the importance of three - electron jastrow terms in the low - density 1d and 2d hegs . improving the jastrow factor for single - determinant backflow wave functions also leads to small improvements in the dmc energy of the 2d heg . this demonstrates the indirect effect that improving the jastrow factor can have on improving the nodal surface , as reported in ref . @xcite . we have made efforts to obtain accurate single - determinant vmc energies for most of the systems studied , but for beh and n@xmath0 we deliberately used inferior one - electron basis sets to see whether we could compensate for this with one - electron jastrow terms . we find that this goal can be achieved by including an @xmath195 jastrow term or anisotropic e n terms , along with the usual @xmath193 term . we thank john trail for producing the hf wave function for h@xmath0 . the calculations were performed on the cambridge high performance computing service facility . in this section we describe our design choices in implementing our jastrow factor in the casino code @xcite . the implementation principles are modularity and extensibility , embracing the flexibility that this jastrow factor has by design rather than impeding it by focusing too strictly on performance . the most important design requirement for modularity is that basis sets be dealt with separately rather than included in the jastrow factor code . basis functions , along with their derivatives when required , are evaluated and stored for later use . for any given term with expansion orders @xmath80 and @xmath81 there are @xmath232 e e and @xmath233 e n functions to evaluate and store . single - electron updates involve recalculating @xmath234 and @xmath235 of these functions . note that the number of basis functions that must be evaluated is independent of the term ranks @xmath10 and @xmath11 . furthermore , it is possible to allow different terms to share basis functions that do not contain optimizable parameters so that , e.g. , the natural powers involved in computing @xmath192 and @xmath193 can be re - used for @xmath194 . a number of properties of the basis functions are required to construct the jastrow factor . equation ( [ eq : lambda_symmetry_invariant_signature ] ) requires knowledge of whether basis functions are symmetric or antisymmetric , and their value , first radial derivative , and angular dependency at the origin are required by eqs . ( [ eq : ee_cusp ] ) , ( [ eq : ee_noaniso ] ) , ( [ eq : ee_nocusp ] ) , ( [ eq : en_cusp ] ) , ( [ eq : en_noaniso ] ) , and ( [ eq : en_nocusp ] ) . the one - contribution , no - contribution , and equal - product constraints in these equations require a table indexing distinct products of two basis functions at any value of their arguments . we implement interfaces that make these properties avaliable so that basis functions can be treated as abstract objects in the construction of the jastrow factor , which makes implementing new basis sets straightforward . cutoff functions are dealt with as additional basis sets with an expansion order of one , and we store information identifying the cutoff functions that are strictly zero , which we use to speed up evaluation of the jastrow factor . for the evaluation of an arbitrary - rank jastrow factor term it is necessary to use efficient procedures that iterate from a given set of electronic and nuclear indices @xmath58 and @xmath59 to the next in a specific order ; explicit loops over scalar integer indices are not an option in static code since the loop depth is variable , and the memory usage of precomputing all possible @xmath58 and @xmath59 scales badly with system size for high ranks @xmath10 and @xmath11 . these procedures should take into account which cutoff functions are zero so that particle sets that do not contribute to the jastrow factor are skipped . efficient handling of localized jastrow factor terms is important because it allows the cost of evaluating a term of any rank to scale linearly with system size if the cutoff lengths are held fixed . we implement a scheme where we construct a list of the electrons that are `` connected '' to each electron and each nucleus via nonzero cutoff functions . for the electronic indices , the value of @xmath236 is iterated between @xmath26 and @xmath2 , then the value of @xmath237 is iterated over the values in the list associated with electron @xmath236 that are greater than @xmath236 , then the value of @xmath238 is iterated over the values in the intersection between the lists associated with @xmath236 and @xmath237 that are greater than @xmath237 , and so on . the procedure that iterates over nuclear indices selects sets of nuclei whose connected - electron lists have nonzero intersections . we iterate over @xmath59 in the outermost loop so that we can feed the intersection of all e n lists into the @xmath58 iterator as an initial list for index @xmath236 . the signature @xmath239 of each group of particles is computed inside the electronic and nuclear loops , identifying the linear parameter channel associated with the group of particles . we then loop over linear parameters in the channel , computing the products of the relevant basis functions which have already been precomputed . in terms without indexing constraints , consecutive linear parameters tend to have very similar expansion indices , i.e. , they multiply most of the same basis functions . in order to save multiplications , it is convenient to buffer partial products so that , e.g. , if only the last two of six expansion indices change from one parameter to the next , we can recover the product of the first four functions from the previous index set and save three of the five multiplications required to combine the six basis functions . in typical qmc calculations individual electron moves , rather than full configuration moves , are proposed , which requires computing an acceptance probability involving the ratio of the trial wave function at the proposed and original positions . to calculate this efficiently one needs to compute the part of the jastrow factor which depends on the position of a single particle @xmath96 , ignoring the contributions not involving @xmath96 . in our implementation we evaluate this one - electron jastrow factor using eq . ( [ eq : gjastrow_sum_of_core ] ) , where we fix @xmath240 and iterate over the rest of @xmath58 . the main difference from the evaluation of the full jastrow factor is that @xmath58 is not sorted , and the permutation required to sort @xmath58 , which amounts to inserting @xmath96 at the correct position in @xmath58 , needs to be taken into account in the presence of antisymmetric e e basis functions , since sign changes will be required by eq . ( [ eq : lambda_symmetry_invariant_signature ] ) . the evaluation of gradients and laplacians of the jastrow factor term can be easily accommodated in the one - electron jastrow evaluation code . for performance reasons it is advisable to implement versions of the evaluation procedure for fixed ranks , with fixed - depth loops that can be optimized by compilers . we implement three optimized versions : one for e e terms , one for e n terms , and one for e e other terms are handled by three generic procedures : one for terms without e e functions , one for terms without e n functions , and one for terms with both e e and e n functions . let us consider the @xmath196 term used for the 2d heg in section [ subsubsec:2dheg ] . this is a system with @xmath241 electrons , half of each spin , and the @xmath196 term corresponds to @xmath242 , @xmath243 , expansion order @xmath244 , and basis functions @xmath245 the spin - pair dependency matrix @xmath28 is of size @xmath30 , but in practice we specify a reduced @xmath246 version of this matrix where each row ( column ) corresponds to different spins . for this system this matrix is @xmath247 where we assign parallel - spin electron pairs a spin - pair dependency index of 1 , and antiparallel - spin electron pairs an index of 2 . the @xmath28 matrix can be regarded as an intrinsic property of the system , but in some cases additional symmetries can be imposed in order to reduce the number of parameters in the jastrow factor term ; for example we would achieve this by setting all elements of @xmath28 to 1 in the 2d heg , choosing to ignore the distinction between parallel- and antiparallel - spin electron pairs . the elements in @xmath28 determine pairwise properties and objects . for example , there are as many sets of nonlinear parameters in a jastrow factor term as different values in @xmath28 ; in this case , there are two cutoff lengths , @xmath248 for parallel - spin electron pairs and @xmath249 for antiparallel - spin electron pairs . four distinct types of three - electron groups can be formed : three up - spin electrons ( @xmath250 ) , two up - spins and one down - spin ( @xmath251 ) , one up - spin and two down - spins ( @xmath252 ) , and three down - spin electrons ( @xmath253 ) . the spin - pair dependency matrices for these four groups are @xmath254 @xmath255 the matrices @xmath256 and @xmath257 are identical , and therefore correspond to a linear parameter channel which is used for groups of three electrons with parallel spins . the matrix @xmath258 can be transformed into @xmath259 via , e.g. , the permutation matrix @xmath260 and therefore both matrices correspond to a second linear parameter channel which is used for groups of three electrons with mixed spins . the signature of the first channel , which we refer to as the @xmath250 channel , is @xmath256 , and the signature of the @xmath251 channel is @xmath259 . both of these matrices are considered sorted by our matrix - sorting algorithm . the symmetry constraints for the parameters in each of the two channels depend on the above matrices . @xmath256 is invariant with respect to the application of any permutation , and for the @xmath250 channel eq . ( [ eq : lambda_symmetry_invariant_signature ] ) equates any two parameters with the same indices in a different order . the signature of the @xmath251 channel is only invariant with respect to one nontrivial permutation , @xmath261 and therefore the symmetry constraints for this channel are @xmath262 we do not allow expansion indices to be zero in the @xmath196 term , and the presence of cutoff functions prevents the `` one contribution '' constraint of eq . ( [ eq : ee_cusp ] ) from being satisfied . it is therefore not possible to impose the kato cusp conditions on the @xmath196 term . this is not a problem in practice because we use this term in conjunction with @xmath192 . since the basis functions are isotropic , the only constraint applicable to this term at coalescence points is eq . ( [ eq : ee_nocusp ] ) for @xmath263 . the derivative of the radial projection of the basis function is @xmath264 } { \partial r } \right]_{r=0 } = \delta_{\nu 2 } - \delta_{\nu 1 } \frac{c}{l_p } \;,\ ] ] and the constraint equation for index @xmath265 is @xmath266 = 0 \;.\ ] ] any two products of pairs of natural powers is equal if the sum of the exponents in each of them is equal . therefore the `` equal - product '' constraint for the @xmath196 term is @xmath267 where @xmath158 ranges from 2 to @xmath268 , and the constraint equation for index @xmath265 thus reduces to for example , @xmath270 contains @xmath271 terms , but if @xmath272 , this operation can be simplified to @xmath273 , which contains @xmath274 terms . in general , the sum of an object which is symmetric with respect to the interchange of any of its @xmath10 indices , where these indices take nonrepeated values , can be rewritten as a sum over sorted sets of indices involving @xmath275 times fewer terms than the original sum . given a symmetric @xmath37 matrix @xmath29 and a rectangular @xmath42 matrix @xmath28 , this algorithm finds the permutation of these matrices , @xmath276 and @xmath277 , that correspond to the lowest collational order according to a predefined sorting criterion . the specifics of such criterion are unimportant ; the only requirement is that for any @xmath37 permutation matrix @xmath60 and @xmath278 permutation matrix @xmath61 , the application of the algorithm to the pair of matrices @xmath279 and @xmath280 also returns @xmath281 and @xmath282 . r. dovesi , v. r. saunders , c. roetti , r. orlando , c. m. zicovich - wilson , f. pascale , b. civalleri , k. doll , n. m. harrison , i. j. bush , p. darco , and m. llunell , _ crystal06 user s manual _ , university of torino ( 2006 ) .	we have developed a flexible framework for constructing jastrow factors which allows for the introduction of terms involving arbitrary numbers of particles . the use of various three- and four - body jastrow terms in quantum monte carlo calculations is investigated , including a four - body van der waals - like term , and anisotropic terms . we have tested these jastrow factors on one- and two - dimensional homogeneous electron gases , the be , b , and o atoms , and the beh , h@xmath0o , n@xmath0 , and h@xmath0 molecules . our optimized jastrow factors retrieve more than 90% of the fixed - node diffusion monte carlo correlation energy in variational monte carlo for each system studied .
multi - output ( mimo ) systems can be designed to provide two types of gains : transmit diversity gain and spatial multiplexing gain@xcite . the full - rate full - diversity space - time block codes ( stbc ) in @xcite can achieve both for 2@xmath12 mimo systems . recently , a fast - decodable full - rate stbc is proposed by s. sezginer , h. sari and e. biglieri @xcite@xcite : @xmath2 \end{split}\ ] ] where @xmath3 with @xmath4 are information symbols , @xmath5 are design coefficients and @xmath6 denotes the complex conjugate . due to its code structure in ( [ bostbc ] ) , @xmath7 has additional zero entries appearing in the upper - triangular matrix after qr decomposition of the equivalent channel matrix , thus making it fast - decodable @xcite@xcite . it is shown in @xcite that the code structure ( [ bostbc ] ) after optimizing for non - vanishing determinant can be rewritten with a single design coefficient . base on this knowledge , in this paper we adopt a simplified version of the code structure ( [ bostbc ] ) by setting @xmath8 to obtain : @xmath9 \end{split}\ ] ] where @xmath10 and @xmath11 is the design coefficient with @xmath12 . our objective is to analytically optimize the design coefficient @xmath13 in ( [ rate2coderot ] ) to enable the full - rate stbc to achieve non - vanishing determinants . in particular , we will consider the influence of different signal constellation topologies , including rectangular quadrature amplitude modulation ( qam ) , amplitude - phase shift keying ( apsk ) and phase shift keying ( psk ) , on the existence of non - vanishing determinants . the rest of this paper is organized as follows . in section [ secdiversity ] , the methods to optimize the design coefficient in ( [ rate2coderot ] ) for both integer - coordinate and non - integer - coordinate signal constellations are described . comparisons of the code in ( [ rate2coderot ] ) with other full - rate codes are shown in section [ secsimulation ] . this paper is concluded in section [ secconclusion ] . in what follows , bold lower case and upper case letters denote vectors and matrices ( sets ) , respectively ; @xmath14 and @xmath15 denote the real and the complex number fields , respectively ; @xmath16 and @xmath17 stand for the real and imaginary parts of a complex element vector and matrix , respectively ; @xmath18^h$ ] denotes the complex conjugate transpose of a matrix ; @xmath19 denotes the determinant of a square matrix . following @xcite , the diversity gain of @xmath20 in ( [ rate2coderot ] ) is denoted as @xmath21 @xcite where the difference matrix @xmath22 , @xmath23 and @xmath24 are stbc matrices based on different information symbols . a full - rank @xmath25 guarantees that @xmath26 is full - rank , and the code @xmath20 in ( [ rate2coderot ] ) will achieve full diversity . when @xmath25 is full rank , the coding gain can be defined as @xmath27\\ & = \underset{\delta \textbf{x}}{\min}\left(\left\| det\left(\delta\textbf{x}\right)\right\|^2 \right ) \end{split}\ ] ] where @xmath28 and @xmath29 are the difference symbols of @xmath30 . @xmath31 can be split into two parts : @xmath32 where @xmath33 @xmath34\\ & ( 1-j ) . \end{split}\ ] ] note that @xmath35 is dependent on the design coefficient @xmath13 , while @xmath36 is decided by the difference symbols only . since @xmath36 is in the form of @xmath37 multiplied by a real number ( determined by specific values of @xmath38 ) , if @xmath39 is plotted on the @xmath40-axis and @xmath41 is plotted on the @xmath42-axis , @xmath36 lies discretely on the line @xmath43 , as shown in fig . [ smc2x2_d1d21 ] and fig . [ smc2x2_d1d22 ] . since @xmath44=\underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}(\|d_1-d_2\|^2)$ ] , @xmath45 is the necessary and sufficient condition for full diversity , and this can be achieved by influencing @xmath35 using the design coefficient @xmath13 . let @xmath46 where @xmath47 and @xmath48 , we have @xmath49+\\ & \left[\left(\| \delta s_3\|^2+\| \delta s_4\|^2\right)v-\left(\| \delta s_1\|^2+\|\delta s_2\|^2\right)u\right]j . \end{split}\ ] ] the coding gain can be analyzed in two different cases as shown below : in this case , @xmath50(1-j).\ ] ] similar to @xmath36 , @xmath35 lies on the line @xmath43 if @xmath51 is plotted on the @xmath40-axis and @xmath52 is plotted on the @xmath42-axis ( as illustrated in fig [ smc2x2_d1d21 ] ) . the discrete loci of @xmath35 on the line @xmath43 depend not only on @xmath38 , but also on @xmath53 and @xmath54 ( the design coefficients ) . let @xmath55 and @xmath56 , from ( [ d_1 ] ) and ( [ d_2 ] ) we get @xmath57 to achieve full diversity gain ( @xmath45 ) , @xmath53 and @xmath54 must be chosen to achieve @xmath58 . note that @xmath59 $ ] due to @xmath48 . hence , in this case the * case i * coding gain is @xmath60\\ = & \underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}\left(\|d_1-d_2\|^2 \right)\\ = & \underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}\left(2\| \tilde d_1-\tilde d_2\|^2\right ) . \end{split}\ ] ] , @xmath36 and @xmath61 of * case i * illustrated in real - imaginary axis graph.,width=259 ] , @xmath36 and @xmath61 of * case ii * illustrated in real - imaginary axis graph.,width=288 ] in this case , we have @xmath62 if @xmath63 , then @xmath64 and @xmath35 will never lie on the line @xmath43 , as shown in fig . [ smc2x2_d1d22 ] . since we have earlier shown that @xmath36 always lies on the line @xmath43 , it implies that @xmath45 , hence full diversity is always achieved by * case ii. as shown in fig . 2 , the euclidean distance between @xmath35 and @xmath36 can be lower bounded by the perpendicular distance between @xmath35 and the line @xmath43 where @xmath36 lies on . [ smc2x2_d1d22 ] . hence , the case ii * coding gain is lower bounded as @xmath65\\ = & \underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}\left(\|d_1-d_2\|^2 \right)\\ \geq&\underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}\left [ \emph{\emph{d}}^2(\emph{\emph{point } } : d_1,~\emph{\emph{line } } : x+y=0)\right ] \\=&\underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}\left[(\| \delta s_3\|^2+\| \delta s_4\|^2-\| \delta s_1\|^2-\| \delta s_2\|^2)^2\right.\\ & \left.(u+v)^2/2\right ] \end{split}\ ] ] where @xmath66 denotes the perpendicular distance from a point to a line . next , we will find the @xmath53 and @xmath54 that satisfy the above full diversity conditions and maximize the coding gain @xmath44 $ ] for the above two cases . to make the optimization process tractable , we will first find the @xmath53 and @xmath54 that maximize the * case i * coding gain ( [ rate2det2 ] ) . then we show that the * case ii * coding gain ( [ rate2det1 ] ) substituted with the @xmath53 and @xmath54 obtained are larger than the maximized ( [ rate2det2 ] ) . hence the * case i * coding gain ( [ rate2det2 ] ) dominates the performance of the code @xmath20 , and the @xmath53 and @xmath54 obtained by maximizing ( [ rate2det2 ] ) will be the global optimum design coefficients . when integer - coordinate signal constellations ( such as rectangular qam ) are applied , the difference symbols also have integer coordinates @xcite@xcite , where the minimum euclidean distance in the signal constellation is fixed at 1 . [ lemma3_conclusion ] _ when integer - coordinate signal constellations are applied , the coding gain ( [ rate2det2 ] ) of the code @xmath20 in * case i * is upper bounded by @xmath67 , and the maximum value can be achieved if and only if @xmath68 . _ in appendix [ proof_theorem1 ] . the following theorem establishes the non - vanishing determinant of @xmath69 with integer - coordinate signal constellations . [ theorem1 ] _ when integer - coordinate signal constellations are applied , the optimum design coefficient @xmath46 to achieve full diversity and maximum non - vanishing coding gain for the code @xmath20 in ( [ rate2coderot ] ) is given by : _ [ rate2det_con11 ] @xmath70 let us first consider * case i. ( [ rate2det_con11 ] ) can be obtained by combining @xmath68 from lemma [ lemma3_conclusion ] and @xmath48 ( by definition ) . next , for case ii * , since @xmath71 and integer - coordinate signal constellations are applied , we have @xmath72 . substituting the @xmath53 and @xmath54 in ( [ rate2det_con11 ] ) to ( [ rate2det1 ] ) , the coding gain becomes @xmath73\\ \geq&\underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min } \left[(\| \delta s_3\|^2+\| \delta s_4\|^2-\| \delta s_1\|^2-\| \delta s_2\|^2)^2\right.\\ & \left.(u+v)^2/2\right]\\ = & 7/8 . \end{split}\ ] ] comparing the * case ii * coding gain expressed in ( [ redet2 ] ) with the * case i * coding gain expressed in lemma [ lemma3_conclusion ] , the * case i * coding gain is lower , hence it is the overall coding gain , and theorem [ theorem1 ] is proved . _ remark _ : the method of proof in this paper , specifically * case i * and * case ii * , are presented in a different way than the proof provided in @xcite . interestingly , however , the optimized design coefficients in both papers are found to be the same . apsk ( amplitude - phase shift keying ) is a high - order modulation scheme commonly used in siso ( single - input single - output ) communications . conventional apsk topology resembles multi - ring psk , or circular qam , as illustrated in fig . [ fig_8apsk ] and fig . [ fig_16apsk ] . compared with rectangular qam , apsk has advantages such as lower constellation peak - to - average - power ratio ( papr ) and robustness against nonlinear distortion in siso communications @xcite . moreover , apsk may lead to larger minimum euclidean distance per unit average power for certain constellation dimension such as 8-apsk @xcite . hence apsk has been adopted by the dvb - s2 standard @xcite . in order for the apsk constellation with arbitrary constellation dimension to achieve non - vanishing coding gain with the code @xmath20 in ( [ rate2coderot ] ) , we may deduce from theorem [ theorem1 ] that : \(1 ) the apsk constellation points should lie on square grids and ring radius @xmath0 ; \(2 ) the design coefficient in ( [ rate2det_con11 ] ) should be adopted for the code @xmath20 . two examples of the proposed apsk constellations are shown in fig . [ fig_8apskm ] and fig . [ fig_16apskm ] . with minimum euclidean distance fixed at 1 , they lead to a non - vanishing coding gain of @xmath67 for the code @xmath20 ( same proof as theorem [ theorem1 ] ) . when non - integer - coordinate signal constellations such as @xmath74-ary phase shift keying ( @xmath74-psk ) are applied , the difference symbols @xmath75 and @xmath76 do not have integer coordinates . this leads to a vanishing determinant for the code @xmath20 in ( [ rate2coderot ] ) even when the minimum euclidean distance is fixed at 1 . the proof is straightforward , hence omitted . although the code @xmath20 in ( [ rate2coderot ] ) with @xmath74-psk constellations has vanishing determinant , the code can still be analytically optimized for a specific constellation size based on the mathematical framework presented earlier . the optimization methodology is described below : step 2 consider * case i * : @xmath77 , whose coding gain expression is shown in ( [ rate2det2 ] ) . given a signal constellation , find out all the values of @xmath78 ; for each value of @xmath78 , find out all the values of @xmath79 . since @xmath80 is a function of @xmath81 , the expression of @xmath82 as a function of @xmath81 can be evaluated . based on these expressions of @xmath82 and @xmath83 $ ] , obtain the maximum value of @xmath84 , and the corresponding @xmath81 . combining the @xmath81 obtained with @xmath48 , we can obtain the corresponding @xmath53 , @xmath54 and the maximized coding gain ; next , consider * case ii * : @xmath85 . substitute the @xmath53 and @xmath54 obtained in step 1 into ( [ rate2det1 ] ) to obtain the * case ii * coding gain . if the * case ii * coding gain is higher than that of * case i * , then the latter is the overall coding gain by definition , and we conclude that the code @xmath20 in ( [ rate2coderot ] ) with design coefficients @xmath46 obtained in * case i * achieves full diversity gain and maximum coding gain . for psk and conventional apsk , this is found to be always true . applying the optimization steps described above to the code @xmath20 in ( [ rate2coderot ] ) with 8-psk constellation , the optimum design coefficients shown in ( [ rate2det_con21 ] ) and the maximum coding gain of @xmath86 are obtained . [ rate2det_con21 ] @xmath87 similarly , the optimized design coefficients and coding gains for the conventional 8-apsk shown in fig . [ fig_8apsk ] and the conventional 16-apsk shown in fig . [ fig_16apsk ] can be found , and are listed in table [ table_apsk ] . they will be used later in fig . [ in the simulations , we assume that the rayleigh fading channel is quasi - static in the sense that the channel coefficients do not change within a codeword , and the channel state information ( csi ) is perfectly known at the receiver . firstly , we show the ml bit error rate ( ber ) performances of the proposed code @xmath20 in ( [ rate2coderot ] ) with the optimized coefficient ( [ rate2det_con11 ] ) , the golden @xcite , pga @xcite , mtd and mcc @xcite codes for 2@xmath12 mimo systems with 4-qam and 16-qam in fig . [ fig_qr21 ] . the ssb code @xcite is equivalent to the proposed code @xmath20 . the results show that the proposed code @xmath20 in ( [ rate2coderot ] ) with design coefficients ( [ rate2det_con11 ] ) has ber performance slightly worse than golden code @xcite , comparable with pga code @xcite , and better than mtd and mcc codes @xcite . on the other hand , as the proposed code structure ( [ rate2coderot ] ) is fast - decodable , it has computational complexity order @xmath88 @xcite , same as the codes in @xcite@xcite . since the computational complexity order of golden code @xcite is @xmath89 , the small performance loss of the proposed code compared to golden code can be viewed as a small penalty to be paid for the complexity reduction . 2 mimo systems with 4-qam and 16-qam constellations.,width=355 ] the ber performance of the code @xmath20 in ( [ rate2coderot ] ) is next compared with other full - rate codes @xcite@xcite@xcite with 8-psk in fig . [ fig_qr22 ] . here the design coefficients in ( [ rate2det_con21 ] ) are adopted for the code * x * in ( [ rate2coderot ] ) , while the optimum design coefficients for the other codes are taken from their respective publications . from the simulation results , we can see that the code * x * in ( [ rate2coderot ] ) achieves a larger ber slope when the snr is high . this is because the other codes , including golden code , were optimized for qam , not psk . 2 mimo systems with 8-psk constellation.,width=367 ] the coding gains of the full - rate stbcs with qam and psk constellations are tabulated in table [ table_codinggain ] with the average power of information symbols normalized to 1 . in all cases , they concur with the ber observations made in fig . [ fig_qr21 ] and [ fig_qr22 ] . 0.8 mm [ cols="^,^,^,^",options="header " , ] comparisons of the properties and performance of the code @xmath20 in ( [ rate2coderot ] ) , when used with the conventional apsk topology versus the proposed apsk topology shown in fig . [ fig_cqam ] , are presented in table [ table_apsk ] and fig . [ 8apsk ] , respectively . note that fig . [ fig_8apsk ] is the best known conventional 8-apsk ( in siso sense ) , while fig . [ fig_16apsk ] is the 16-apsk adopted by the dvb - s2 standard @xcite . in the ber simulations , the corresponding optimum code design coefficients @xmath13s from table [ table_apsk ] are applied . note from table [ table_apsk ] that the proposed apsk does not need to change its design coefficient @xmath13 for different constellation dimensions as it achieves non - vanishing determinant , but this is not true for the conventional apsk . interestingly , table [ table_apsk ] shows that although the proposed apsk shown in fig . [ fig_8apskm ] and fig . [ fig_16apskm ] have smaller minimum euclidean distance ( hence lower papr for the proposed 8-apsk ) , they achieve higher coding gain than the conventional apsk . this is because the coding gains do not depend linearly nor solely on the minimum euclidean distance , as shown in ( [ rate2det2 ] ) and ( [ rate2det1 ] ) . [ 8apsk ] shows that the code @xmath20 in ( [ rate2coderot ] ) with the proposed 8-apsk has much better performance than the conventional 8-apsk , while the proposed 16-apsk has similar performance as the conventional 16-apsk at high snr . [ 8apsk ] also testifies that the code design coefficients shown in table [ table_apsk ] for the conventional 8/16 apsk achieve full diversity . 0.8 mm ' '' '' [ 0pt]apsk & euclidean dis . & coefficient@xmath90 : @xmath13 & [ 0pt]coding gain + conventional & & & + 8-apsk @xcite & [ 0pt]0.9194 & [ 0pt]@xmath91 & [ 0pt]0.0230 + proposed & & & + 8-apsk & [ 0pt]0.8165 & [ 0pt]@xmath92 & [ 0pt]0.2222 + conventional & & & + 16-apsk @xcite & [ 0pt]0.5848 & [ 0pt]@xmath93 & [ 0pt]0.0004 + proposed & & & + 16-apsk & [ 0pt]0.5 & [ 0pt]@xmath92 & [ 0pt]0.03125 + in ( [ rate2coderot ] ) in 2@xmath12 mimo systems with the conventional and proposed 8/16-apsk constellations shown in fig . [ fig_cqam].,width=355 ] [ proof_theorem1 ] let us first introduce lemma [ lemma1_conclusion ] and lemma [ lemma2_conclusion ] which will be used later to prove lemma [ lemma2_conclusion ] and lemma [ lemma3_conclusion ] , respectively . in the following , @xmath97 denotes that @xmath98 divides @xmath99 , and @xmath100 denotes that @xmath98 can not divide @xmath99 . as the value of @xmath109 is equal to 0 or 1 for any integer @xmath99 and @xmath110 , one of the following two equations must hold @xmath111 @xmath112 in other words , @xmath113 , @xmath114 , @xmath115 , @xmath116 must hold at the same time , or @xmath117 , @xmath118 , @xmath119 , @xmath120 must hold at the same time . : since @xmath121 , we have @xmath122 . as the value of @xmath123 is equal to 1 for any odd integer @xmath99 , it follows from @xmath122 that @xmath124 are even integers , i.e. , @xmath125 , @xmath126 , @xmath127 and @xmath128 . then , @xmath129 where @xmath130 , @xmath131 , @xmath132 and @xmath133 are integers . applying the conclusions in case @xmath134 , we have @xmath135 , @xmath136 , @xmath137 , @xmath138 at the same time , or @xmath139 , @xmath140 , @xmath141 , @xmath142 at the same time . since @xmath102 , @xmath144 , @xmath145 , @xmath146 and @xmath147 , we have @xmath148 where @xmath149 , @xmath150 , @xmath151 and @xmath152 are integers . following the conclusions in case @xmath153 , it can be shown that @xmath149 , @xmath150 , @xmath151 and @xmath152 are even integers , i.e. , @xmath154 , @xmath155 , @xmath156 and @xmath157 . then , @xmath158 where @xmath159 , @xmath160 , @xmath161 and @xmath162 are integers . applying the conclusions in case @xmath134 , we have @xmath163 , @xmath164 , @xmath165 , @xmath166 at the same time , or @xmath167 , @xmath168 , @xmath169 , @xmath170 at the same time . hence , @xmath171 , @xmath172 , @xmath173 , @xmath174 at the same time , or @xmath175 , @xmath176 , @xmath177 , @xmath178 at the same time . \2 ) when @xmath194 , @xmath195 at the same time , then @xmath197 and @xmath198 where @xmath199 and @xmath200 are odd integers . thus , @xmath201\emph{\emph { mod } } 2^k=2^{k-1}[(m_1+m_2)\emph{\emph { mod } } 2]=0 $ ] , i.e. , @xmath196 . in * case i * , @xmath203 . for integer - coordinate signal constellations , the difference symbols can be denoted as @xmath204 where @xmath179 are integers and will not be zeros at the same time , i.e. , @xmath205 . hence , we have let @xmath207 be expressed as @xmath208 where @xmath104 is a non - negative integer and @xmath209 is an odd integer . following lemma [ lemma2_conclusion ] , it can be shown that @xmath210 , i.e. , @xmath211 . hence , @xmath212 where @xmath213 is an integer and we have @xmath214 since @xmath213 is an integer decided by @xmath79 and @xmath209 is an odd integer , we have @xmath215 . the equality holds if and only if @xmath216 . since the coding gain of @xmath20 in * case i * is @xmath44=\underset{\delta s_1\emph{\emph { to } } \delta s_4}{\min}(2\|\tilde d_1-\tilde d_2\|^2)$ ] , it is easy to see that @xmath44\leq1/2 $ ] and the equality holds if and only if @xmath216 . j. paredes , a. b. gershman , and m. g. alkhanari , a new full - rate full - diversity space - time block code with nonvanishing determinants and simplified maximum - likelihood decoding , _ ieee trans . signal process . 2461 - 2469 , june 2008 . v. tarokh , n. seshadri , and a. r. calderbank , space - time codes for high data rate wireless communication : performance criterion and code construction , _ ieee trans . inf . theory _ , 744 - 765 , mar . 1998 . etsi tr 102 376 v1.1.1 ( 2005 - 02 ) , digital video broadcasting ( dvb)-user guidelines for the second generation system for broadcasting , interactive services , news gathering and other broadband satellite applications ( dvb - s2 ) , 2005 .	full - rate stbc ( space - time block codes ) with non - vanishing determinants achieve the optimal diversity - multiplexing tradeoff but incur high decoding complexity . to permit fast decoding , sezginer , sari and biglieri proposed an stbc structure with special qr decomposition characteristics . in this paper , we adopt a simplified form of this fast - decodable code structure and present a new way to optimize the code analytically . we show that the signal constellation topology ( such as qam , apsk , or psk ) has a critical impact on the existence of non - vanishing determinants of the full - rate stbc . in particular , we show for the first time that , in order for apsk - stbc to achieve non - vanishing determinant , an apsk constellation topology with constellation points lying on square grid and ring radius @xmath0 needs to be used . for signal constellations with vanishing determinants , we present a methodology to analytically optimize the full - rate stbc at specific constellation dimension .
survival techniques are useful in situations where regular regression procedures are inadequate . for instance , where the probability of survival past a certain time is of more interest than the expected time of survival , survival techniques are more appropriate . the same holds true when there are censored data . censored data may arise due to partial follow - up ( until a specific time ) , loss to follow - up , or a subject dropping out of a study . attrition is quite common in medical studies where a subject is followed to a particular event of interest . in such cases , the status of these subjects with respect to the event of interest is only known up to a particular time point . competing risks data often arise in survival and reliability analyses . these multiple cause data represent time to failure ( occurrence of event of interest ) and can be due to a host of causes . in a cancer study , for example , a significant number of deaths due to causes other than cancer may be expected post - treatment because the average age of cancer patients is high @xcite . some existing methods for dealing with failure time data include parametric mixture models based on failure time densities @xcite , proportional hazards models @xcite , cause - specific hazard functions @xcite , accelerated failure time models @xcite , first hitting time models @xcite , frailty models @xcite , and so on . a classical mixture model for competing risks adopted a proportional hazards model with piecewise constant baseline hazards @xcite . a traditional approach to competing risks is the latent failure time approach @xcite which makes untestable assumptions about the independence of the competing risks ; this is unlike the mixture approach @xcite . we present a novel method to account for censored competing risks data using a mixture of accelerated failure time models in a cluster - weighted modelling ( cwm ) framework . this approach can be called the cluster - weighted accelerated failure time ( cwaft ) model . section [ cwm ] gives an introduction to cluster - weighted modelling . section [ methods ] briefly goes over accelerated failure time models and the mathematical details involved for cwaft models . in section [ results ] , we present our results on some simulated and real censored data , where the time to failure and cause of failure were only recorded for some of the observations . finally , we conclude with a discussion and some ideas for further research . @xcite give a good introduction to mixture models for the analysis of failure time data . in survival analysis , mixture models can be defined in terms of the density function @xmath0 or in terms of the survival function @xmath1 of @xmath2 where @xmath2 , is the ( non - negative , continuous ) random variable representing the time to the event of interest . hence , @xmath3 or @xmath4 , where @xmath5 and @xmath6 are the @xmath7th component density and the @xmath7th component survival function , respectively . @xcite introduced cwm in a more general statistical context . this methodology allowed for the decomposition and modelling of the joint probability of a response variable and a set of explanatory variables . let @xmath8 be the pair of random vector @xmath9 and random variable @xmath10 defined on @xmath11 with joint probability distribution @xmath12 . here , the response variable @xmath10 has values in @xmath13 and the explanatory variable @xmath9 is a @xmath14-dimensional vector with values in @xmath15 . if @xmath11 can be partitioned into @xmath16 disjoint groups , such that @xmath17 , the joint probability @xmath12 can be decomposed as @xmath18 where @xmath19 is the conditional density of @xmath10 given @xmath20 and @xmath21 , @xmath22 is the probability density of @xmath9 given @xmath21 , , and @xmath23 is the mixing weight , where @xmath24 ( @xmath25 ) and @xmath26 . @xmath27 denotes the set of all parameters . here , @xmath23 is the mixing weight of @xmath21 , where @xmath24 ( @xmath25 ) and @xmath26 . the mixing proportions @xmath28 in survival mixture models are often assumed to follow a logistic model of the covariates @xcite . using a cwm framework , the model presented in this paper is not necessarily bound by this restriction . here , @xmath19 can be viewed as weighted by both the marginal @xmath22 and unrestricted mixing weights @xmath28 . accelerated failure time ( aft ) models are a popular parametric regression alternative to proportional hazards regression . some common aft models employ the log - normal , log - logistic , and weibull distributions @xcite . @xmath29 is modeled analogous to classical regression . for a log - normal distribution , @xmath30 where @xmath31 is the mean and @xmath32 is the standard deviation of @xmath33 . this is @xmath34 where @xmath35 is the survival function of the standard gaussian distribution . a regression model can be constructed for @xmath10 given @xmath20 with @xmath36 with covariates , the linear model representation is @xmath37 , where @xmath38 is the error distribution ( log - scale form ) and @xmath20 provides the covariates . if a standard gaussian distribution is used for the error distribution , this results in a log - normal aft model . the underlying density and survival function for @xmath10 are @xmath39 and @xmath40 here , @xmath41 and because @xmath42 , where @xmath43 , @xmath44 . we propose that @xmath45 in the cwm joint likelihood decomposition can be modeled as a gaussian distribution where the random variable @xmath10 is @xmath33 , where @xmath2 , the time to event , follows a log - normal distribution . recall that cwm decomposes the joint probability @xmath12 as follows : @xmath46 the observed data likelihood is then @xmath47^{l_{ig } } \times \\ & \qquad \prod_{i=1}^{c } \sum_{g=1}^g s_{y}(y_{ig}\|{\mathbf{x}}_i , { \mbox{\boldmath $ \chi$}}_g)\phi_d({\mathbf{x}}_i\| { \mbox{\boldmath $ \psi$}}_g ) \pi_g,\end{aligned}\ ] ] where @xmath48 is 1 if observation @xmath49 is known to have failed from cause @xmath7 and 0 otherwise , @xmath50 , and @xmath51 . @xmath52 and @xmath53 are the number of censored and total observations , respectively . the expectation - maximization framework is a popular technique for parameter estimation @xcite . this involves maximizing the expected complete - data log - likelihood . we use the alternating expectation conditional maximization ( aecm ) algorithm @xcite , which is a variation of the em algorithm that allows for different complete data at different stages . the em algorithm was coded in r @xcite . let @xmath54 be a sample of @xmath53 independent observations . then , the corresponding complete - data likelihood function can be written in the form @xmath55^{z_{ig}}.\ ] ] here , two latent variables are introduced . if @xmath57 comes from the @xmath7th population and @xmath58 otherwise . this corresponds to the conditional probability that individual @xmath49 will eventually fail from risk @xmath7 , conditional on no failure having occured by time @xmath59 . also , @xmath60 represents the uncensored failure time conditional on the @xmath61th individual failing from the @xmath7th cause ( @xmath7th component ) . for uncensored individuals , @xmath62 is observed ; that is , it is merely @xmath63 . as an illustration , consider @xmath64 and assume that @xmath65 corresponds to the competing risk and @xmath64 corresponds to the cause of interest . then , @xmath66 represents the failure time associated with the censored time @xmath63 conditional on the @xmath49th individual failing from the competing cause . using the accelerated failure time framework , the complete - data log - likelihood can be decomposed as @xmath67 \\ & = { \mathcal{l}}_{1c } ( { \mbox{\boldmath $ \chi$ } } ) + { \mathcal{l}}_{2c } ( { \mbox{\boldmath $ \psi$ } } ) + { \mathcal{l}}_{3c } ( { \mbox{\boldmath $ \pi$ } } ) , \end{aligned}\ ] ] where @xmath68 ^ 2}{\sigma_{g}^{2}}\right ] \\ { \mathcal{l}}_{2c } ( { \mbox{\boldmath $ \psi$ } } ) & = \frac{1}{2 } \sum_{i=1}^n \sum_{g=1}^g z_{ig } \left[- p \log 2\pi - \log \|{\mbox{\boldmath $ \sigma$}}_g\| - ( { \mathbf{x}}_i-{\mbox{\boldmath $ \mu$}}_g ) ' { \mbox{\boldmath $ \sigma$}}_g^{-1 } ( { \mathbf{x}}_i-{\mbox{\boldmath $ \mu$}}_g)\right ] \label{l2c } \\ { \mathcal{l}}_{3c } ( { \mbox{\boldmath $ \pi$ } } ) & = \sum_{i=1}^n \sum_{g=1}^g z_{ig } [ \log \pi_g ] .\end{aligned}\ ] ] given the parameters @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 at the @xmath74th iteration , the expected complete - data log - likelihood is @xmath75 \notag \\ & = \sum_{i=1}^n \sum_{g=1}^g \tau_{ig}^{(k ) } [ q_{1}({\mbox{\boldmath $ \chi$}}_g\|{\mbox{\boldmath $ \theta$}}^{(k ) } ) + q_{2}({\mbox{\boldmath $ \psi$}}_g\|{\mbox{\boldmath $ \theta$}}^{(k ) } ) + \log \pi_g^{(k ) } ] , \end{aligned}\ ] ] where @xmath76 provides the current value on the @xmath74-iteration for the censored observations and @xmath77 ^ 2}{\sigma_{g}^{2}}\right ) \right ] , \\ q_{2}({\mbox{\boldmath $ \psi$}}_g\|{\mbox{\boldmath $ \theta$}}^{(k ) } ) & = \frac{1}{2 } \left[- p \log 2\pi - \log \|{\mbox{\boldmath $ \sigma$}}_g\| - ( { \mathbf{x}}_i-{\mbox{\boldmath $ \mu$}}_g ) ' { \mbox{\boldmath $ \sigma$}}_g^{-1 } ( { \mathbf{x}}_i-{\mbox{\boldmath $ \mu$}}_g)\right].\end{aligned}\ ] ] here , for an observed failure time , @xmath78 . for right censored times @xmath79 ( associated response variables are @xmath80 ) , following @xcite we observe that the conditional probability density function of an observation censored to the right is that of the random variable @xmath59 truncated to the left at @xmath81 . hence , @xmath82 where @xmath83 in which @xmath84 is the lifetime that would have been observed if @xmath85 has not been censored ; that is , if it had been observed . then , the conditional expected value @xmath86 for a right censored time is @xmath87 where @xmath88 . note that this is also the mean of the truncated normal distribution ( see appendix [ appeyderive ] ) . similarly , @xmath89 is ( see appendix [ appey2derive ] ) @xmath90 + \\ & \qquad 2 \mu^{(k)}_{g(y ) } { \mathbb{e}}(y_{ig } ) - \mu^{2(k)}_{g(y)}. \end{aligned}\ ] ] the m - step requires the maximization of the conditional expectation of the complete - data log - likelihood with respect to @xmath27 . the updates for the parameters @xmath91 , @xmath92 , and @xmath93 can be found by taking the derivative with respect to the appropriate parameter @xmath94 these closed form updates can be found in @xcite . the updates @xmath95 , and @xmath96 can similarly be derived ( appendix [ mstepapp ] ) : @xmath97 } { \sum_{i=1}^n \tau_{ig}^{(k ) } } .\end{aligned}\ ] ] aitken s stopping criterion was used to determine convergence of the algorithm . aitken s acceleration at iteration @xmath74 can be calculated as @xmath98 where @xmath99 , @xmath100 , and @xmath101 are the log - likelihood values from iterations @xmath102 , @xmath74 , and @xmath103 , respectively . this can be used to compute the asymptotic estimate of the log - likelihood at iteration @xmath103 @xcite : @xmath104 the em algorithm is stopped when @xmath105 @xcite . we use overall survival functions and cumulative incidence function for judging model fits . the overall survival function is estimated in the fashion of @xcite by averaging individual estimates of subjects , and compared to a nonparametric estimate by way of a kaplan meier curve @xcite . this statistic does not utilize information on cause of death or covariates . the overall survival function is @xmath106 . since kaplan - meier ( km ) curves are not appropriate for the event of interest or the competing event directly , cumulative incidence curves are used . the cumulative incidence function can be defined as the cause - specific failure probability up to a certain time point @xmath59 . the cumulative incidence curves are also presented with fits to the nelson - aalen cumulative incidence estimator @xcite . following @xcite , the cumulative incidence function for the @xmath7th type of failure is @xmath107 . this was estimated by @xmath108 . the standard errors were calculated using a non - parametric bootstrap @xcite that was adjusted to account for the competing risks structure as in @xcite . as before , let @xmath52 be the number of censored observations . then , let @xmath109 @xmath110 be the number of observed failures due to the @xmath7th cause . each bootstrap sample was obtained by sampling with replacement separately from each of the @xmath111 sets with the size of each bootstrap subsample equal to @xmath109 and @xmath52 , respectively . the data @xmath57 were re - sampled 100 times independently . the standard deviation of the resulting bootstrap maximum likelihood estimates yield an approximation of the standard error of the estimates . note that we present results for a simulation with two covariates here for simplicity ; however , the model is not bound by this restriction and seems to perform quite well in simulations with more than two covariates . the specifications for the gaussian covariate vectors for group 1 were mean@xmath112 and covariance @xmath113 . the specifications for the gaussian covariate vectors for group 2 were mean@xmath114 and covariance @xmath115 . the regression intercepts used for the groups were @xmath116 and @xmath117 , respectively . the regression coefficients were @xmath118 and @xmath119 , respectively , and the error @xmath120 . fifty observations were censored ( type ii right , noninformative ) in total by subtracting normally distributed values from randomly selected observations from both groups . + .parameter estimates and standard errors ( rounded off to 2 decimals ) for simulated data . [ cols="<,<,<,<,<",options="header " , ] parameter estimates with standard errors in parenthesis , where @xmath7=1 and 2 refer to the cause of interest and the competing cause , respectively . + table [ parest4d ] contains the parameter estimates and standard errors . apart from the high percentage of censoring , this dataset is also unique in that censoring begins relatively late ( figure [ real2osf ] ) . the estimated cumulative incidence functions are still able to capture the trend of the non - parameteric estimators quite closely ( figure [ real2cuminc ] ) . a novel approach to classifying censored data from different groups while utilizing the distribution of the covariates was presented . in simulated and real data sets , the algorithm showed good performance and was able to extract marginal probabilistic behaviour quite well . the algorithm is quite stable as shown in regards to initialization for the em . sometimes , a reasonable number was substituted where nan or infinity errors were encountered while calculating @xmath86 or @xmath89 . commonly , in the literature , the distribution of the covariates is taken into account to alleviate the impact of any mis - recorded covariates , unobserved heterogeneity , or in the presence of evidence that the data are from particular subsamples of the population . these usually correspond to data misspecification ( or missing data ) models , or random effect models . in survival literature , frailty models attempt to account for unobserved heterogeneity by including a random effect . here , the marginal distribution of the covariates @xmath121 is directly taken into account in the maximization of the likelihood function in the em algorithm . this is also novel in the field of survival analysis to the authors knowledge . note that for both real data sets , the competing risk was death from other events . also note that it is not always easy to distinguish between a cured individual and a susceptible individual with a large failure time . @xcite gives a good discussion on how care must be exercised in assuming that a fraction of patients is cured . having said that , a cure rate can estimated by following @xcite . assuming that a patient is cured if the patient dies from a competing cause at a time @xmath122 ( for example , @xmath123 years has been used in breast cancer studies ) without any symptoms of the disease , then from our model , the cure rate can be calculated as @xmath124 . here , @xmath125 is the proportion of individuals who fail from one of the competing causes , that is , will not fail from the cause of interest . @xmath126 refers to the conditional survival function for failure from a competing cause . note that here @xmath125 is being adjusted by excluding those patients with death times smaller than @xmath122 as being cured to account for early deaths that may have been due to postoperative complications and not because they did not have any symptoms of the cause of interest . because we are fitting a gaussian distribution to the marginal distribution of the covariates , currently only continuous covariates can be used . a convenient way to include discrete covariates might be to use them as concomitant variables in a logistic model of the covariates for the probability of failure from a particular risk . however , more work is required to eliminate this aspect from the methodology . incorporating something akin to a latent trait model for the covariates might be fruitful , for instance if we have data on a few levels of dosage . we used the log - normal distribution here , but other commonly used distributions like log - logistic , weibull , generalized gamma , etc . could also be used depending on the data . furthermore , semi - parametric extensions of the model might lead to more flexible alternatives . due to the nature of the likelihood formulation , dealing with missing data in the covariates might also be worthwhile . this work is supported by a alexander graham bell canada graduate scholarship ( cgs - d ) and discovery grant from the natural sciences and engineering research council ( nserc ) of canada . @xmath128 let @xmath129 . then , @xmath130 and @xmath131 , \\ & = \mu + \sigma \frac{\phi\left(\frac{y^-\mu}{\sigma}\right)}{1-\phi\left(\frac{y^-\mu}{\sigma}\right)}. \\\end{aligned}\ ] ] @xmath133 where @xmath134 . again , let @xmath129 . hence , @xmath135 where @xmath136 . now , using integration by parts , @xmath137 + \\ & \qquad 2 b^ * \mu{\mathbb{e}}^{(k)}(y)\left(1-\phi\left(\frac{y^-\mu}{\sigma}\right)\right ) -b^\mu^2\left(1-\phi\left(\frac{y^-\mu}{\sigma}\right)\right ) , \\ & = \sigma_g^2\left[\frac{\left(\frac{y^-\mu_g}{\sigma_g}\right)\phi{\left(\frac{y^-\mu_g}{\sigma_g}\right ) } + 1 - \phi\left(\frac{y^-\mu_g}{\sigma_g}\right)}{1-\phi\left(\frac{y^*-\mu_g}{\sigma_g}\right)}\right ] + 2\mu{\mathbb{e}}^{(k)}(y_{ig } ) -\mu^{2}_{g } . \\\end{aligned}\ ] ] for @xmath138 @xmath139 , @xmath140 yields @xmath141 & = 0 \\ \sum_{i=1}^n \tau_{ig}^{(k ) } \left({\mathbb{e}}^{(k)}(y_{ig})-{\mathbf{b}}_g^{'(k)}{\mathbf{x}}_i\right ) & = b_{0g}^{(k ) } \sum_{i=1}^n \tau_{ig}^{(k)},\end{aligned}\ ] ] and then we get @xmath142 similarly , for @xmath143 @xmath139 , @xmath144 implies @xmath145{\mathbf{x}}'_n = \boldsymbol{0 } ' , \ ] ] yielding @xmath146 for @xmath147 @xmath148 ) , @xmath149 yields @xmath150 this implies @xmath151 bhning d , dietz e , schaub r , schlattmann p , lindsay bg ( 1994 ) the distribution of the likelihood ratio for mixtures of densities from the one - parameter exponential family . annals of the institute of statistical mathematics 46(2):373388 yamaguchi k ( 1992 ) accelerated failure - time regression models with a regression model of surviving fraction : an application to the analysis of permanent employment in japan . journal of the american statistical association 87(418):284292	a novel approach for dealing with censored competing risks regression data is proposed . this is implemented by a mixture of accelerated failure time ( aft ) models for a competing risks scenario within a cluster - weighted modelling ( cwm ) framework . specifically , we make use of the log - normal aft model here but any commonly used aft model can be utilized . the alternating expectation conditional maximization algorithm ( aecm ) is used for parameter estimation and bootstrapping for standard error estimation . finally , we present our results on some simulated and real competing risks data .
the physical properties of interfaces between complex oxides are of great current interest @xcite . a prominent example is the interface between the two band insulators @xmath5 and @xmath6 at which highly mobile carriers are confined and give rise to electronic and superconducting phenomena that can be tuned with a gate voltage @xcite . another interesting example involves the magnetic proximity effect ( mpe ) at the interface between the cuprate high-@xmath0 superconductor ( sc ) @xmath1 ( ybco ) and ferromagnet ( fm ) @xmath2 ( lcmo ) @xcite . with polarized neutron reflectometry ( pnr ) it was found that , in the vicinity of the interface , the fm order of the mn moments is strongly suppressed @xcite . this phenomenon has been discussed in terms of a _ _ or a _ _ . in addition , it was shown with x - ray magnetic circular dichroism ( xmcd ) that the interfacial cu ions acquire a ferromagnetic moment of about 0.2@xmath7 which is antiparallel to the one of mn @xcite . recent x - ray resonant magnetic reflectometry ( xrmr ) studies have demonstrated that these cu moments reside in the interfacial cuo@xmath3 planes @xcite . furthermore , x - ray linear dichroism ( xld ) studies revealed that the interfacial cu ions undergo an orbital reconstruction which yields a large hole density in the cu @xmath4 orbitals ( that are almost fully occupied in the bulk ) @xcite . both the orbital reconstruction and the magnetic moment of the interfacial cu ions have been explained in terms of a strong hybridization between the cu and mn @xmath4 orbitals which leads to an antiferromagnetic exchange interaction ( af - ei ) between the cu and mn spins @xcite . in this context , the ferromagnetic cu moment is induced by the af - ei with the mn moments , and one expects that the magnitude of this cu moment is proportional to the strength of this coupling @xcite . we have performed xmcd , xld and pnr measurements on a series of ybco / lcmo multilayers ( mls ) for which the strength of the af - ei between cu and mn has been altered by changing the electronic and magnetic properties of the lcmo layers . as described in the experimental section , this goal has been achieved by changing the growth and annealing conditions , as well as the thickness of the lcmo layers . to our surprise , we have found that the magnitude of the fm cu moment is almost independent of the strength of the af - ei with mn , i.e. , it remains sizeable if the af - ei is strongly reduced or even absent . this suggests that the fm order of the cu moments is not induced by the coupling to the fm mn moments of lcmo or by a transfer of spin - polarized charge carriers . instead , this cu moment seems to be intrinsic to the interfacial cuo@xmath3 planes . this circumstance is discussed in terms of a fm intra - planar cu - cu exchange interaction that is brought about by the partially occupied cu @xmath4 orbitals . [ pld ] the ybco / lcmo ( yl ) mls have been grown on @xmath8-oriented @xmath9 ( lsat ) substrates using the pulsed laser deposition ( pld ) technique . the _ layer - by - layer _ growth mode and the overall layer thickness have been monitored with _ in - situ _ reflection high energy electron diffraction ( rheed ) as described in ref . three different kinds of mls , in the following we denote them as @xmath10 , @xmath11 and @xmath12 , have been prepared for which the lcmo layers have substantially different electronic and magnetic properties . this has been achieved by changing the growth and annealing conditions as to alter the concentration of oxygen and cation vacancies of the lcmo layers . for all samples , we preheated the lsat substrates to 825@xmath13c in @xmath14 mbar of o@xmath3 for 30 minutes prior to the depositions in order to cure the mechanically polished surface . the samples of type @xmath10 have been grown with 10 bilayer ( bl ) repetitions following a similar procedure as described in ref . @xcite . the ybco layers with a thickness of @xmath1510 nm and the lcmo layers with @xmath1510 nm were grown with an oxygen partial pressure of 0.34 mbar , a laser fluence of @xmath16j/@xmath17 , and a repetition rate of @xmath18hz . after deposition , they were cooled to 700@xmath13c at a rate of 10@xmath13c / min while the oxygen partial pressure was gradually increased to @xmath19bar . subsequently , the samples were further cooled to 485@xmath13c at a rate of 30@xmath13c / min where they were kept for one hour . the temperature was then slowly decreased to room temperature before removing them from the pld chamber . to ensure a full oxygenation of the ybco layers , we performed an _ ex - situ _ annealing at 485@xmath13c in a continuous flow of o@xmath3 for 12 hours . the samples of type @xmath11 have nominally the same ybco and lcmo layer thicknesses as @xmath10 , and have been grown with 1 , 6 and 10 bl repetitions . these samples have been protected with a capping layer of about 1.5 nm @xmath5 ( lao ) . different o@xmath3 partial pressures of 0.34 mbar and 0.12 mbar have been used during the growth of the ybco and lcmo layers , respectively . the laser fluence was kept at 2.0j/@xmath20 and the repetition rate at 2hz . the following _ in - situ _ cooling and annealing procedures were the same as for @xmath10 , except for a lower cooling rate of only 10@xmath13c / min ( instead of 30@xmath13c / min for @xmath10 ) to 485@xmath13c . no post annealing was performed since the ybco layers grown at this lower laser repetition rate have already rather high @xmath0 values ( see subsection [ rt - mr - mt - mh ] ) . sample @xmath12 is a single bl with about 19 nm of ybco and 5 nm of lcmo . it is also protected with a lao capping layer . it was grown with the same pld parameters as @xmath11 , the only difference concerns the _ in - situ _ annealing procedure for which the cooling from 700@xmath13c to 485@xmath13c was done at a faster rate of 30@xmath13c / min . no _ ex - situ _ annealing was performed to this sample . [ xrd - rbs ] the structural quality of all the mls has been confirmed with _ in - situ _ rheed and _ ex - situ _ x - ray diffraction ( xrd ) . the latter also demonstrates the epitaxial growth of the layers with the c - axis of ybco oriented along the surface normal . representative xrd patterns for the samples of type @xmath10 can be found in ref . @xcite and for @xmath11 in figure [ fig - xrd ] ( a ) . the fitting of the x - ray reflectivity profiles of the @xmath11 samples with 1 and 10 bl repetitions , shown in figure [ fig - xrd ] ( b ) , was performed with the genx software @xcite . it yields a layer thickness of about 9.7 nm for ybco and 9.1 nm for lcmo . the roughness of the ybco / lcmo interfaces are 6 and 10 for the samples with 1 and 10 bl repetitions , respectively . ( a ) symmetric @xmath21-@xmath22 x - ray diffraction curve of @xmath11 ( with 10 bl repetitions ) along the [ 00l ] direction . thickness oscillations around the high intensity bragg peaks testify for the quality of the ml . ( b ) x - ray reflectivity profiles ( symbols ) for @xmath11 type samples with 1 and 10 bl repetitions . the solid red lines are the best fits to the data . ] cross - section high - resolution scanning transmission electron microscopy ( stem ) observations of a @xmath11 type sample were carried out in an aberration - corrected jeol jem - arm200cf , operated at 200kv and equipped with a cold field emission gun and a gatan quantum electron energy - loss spectrometer ( eels ) . the convergence semi - angle was around 35mrad , while the collection semi - angle was 28mrad , approximately . the specimens were prepared by conventional methods of grinding and ar - ion milling . random noise in the eels data was removed by means of principal component analysis @xcite . eels elemental mapping was performed by integrating the signals under the characteristic elemental edges after background subtraction using a power law . the integration windows were typically around @xmath23-@xmath24ev wide . stem - eels images show coherent , epitaxial interfaces ( see figure [ fig - tem - eels ] , left panel ) . occasional defects are present such as double cuo chain layers or one unit cell interface steps giving rise to anti phase boundaries , all of these being typical defects observed in ybco . the cuo chain layers are easily identifiable in the high resolution z - contrast images due to their slightly darker contrast @xcite . in most cases , the interfaces are found to be symmetric . regardless of top or bottom relative positions , the dominant atomic plane stacking found is such that a manganite mno@xmath3 plane faces a cuprate bao plane , as shown by the eels profiles such as the ones in figure [ fig - tem - eels ] ( right panel ) . left panel : atomic resolution , z - contrast high angle annular dark field image of the ybco / lcmo / ybco stacking in a @xmath11 type sample . right panel : elemental maps obtained from the analysis of the mn @xmath25 ( red ) , ba @xmath26 ( blue ) and la m@xmath27 ( green absorption edges ) . an rgb overlay of the three maps , along with a line profile on a matching color scale of the normalized integrated intensities is also shown . some spatial drift is visible . ] rutherford backscattering ( rbs ) measurements were performed at the 6mv tandem accelerator of the laboratory of ion beam physics at eth zurich using a 2mev @xmath28he beam and a silicon pin diode detector under a backscattering angle of 168@xmath29 @xcite . the experimental data have been analyzed by the rump code @xcite . we measured two different types of lcmo films that were grown similar to the lcmo layers in @xmath10 and @xmath11 . to be most suitable for the rbs measurements , the lcmo films were 100 nm thick and grown on mgo substrates . the obtained stoichiometry of the lcmo samples is listed in tab . [ rbs - stoi ] . whereas the lcmo layers in @xmath10 are more or less stoichiometric , in @xmath11 there is a significant deficiency of oxygen and a deviation of the cation content from the nominal value . accordingly , the lcmo layers in @xmath11 are likely to have a somewhat reduced hole content . note that the uncertainty of the content of the heavier elements , like la , ca , and mn , is 1 - 3% , whereas for oxygen it is up to 5% . .stoichiometry of 100 nm thick lcmo films grown under similar conditions as @xmath10 and @xmath11.[rbs - stoi ] [ cols="<,<,<,<,<",options="header " , ] [ rt - mr - mt - mh ] the dc magnetization has been measured using the vibrating sample magnetometer ( vsm ) option of a physical property measurement system ( ppms ) from quantum design . figure [ fig - mt - vsm ] displays the temperature and field dependent dc magnetizations ( m - t and m - h curves ) of the samples yl_1 - 3 . the m - t data were acquired during field cooling at a rate of 2k / min in 0.1 t applied parallel to the sample surface . the m - h loops at 80k were recorded after cooling the samples in 9 t . the lcmo layers in @xmath10 have a curie temperature of @xmath30@xmath31 k and a sizeable magnetization of 2.0@xmath7/mn . as compared to @xmath10 , the lcmo layers in @xmath11 have a noticeably lower curie temperature of @xmath30@xmath32k and a somewhat higher magnetization of 2.5@xmath7/mn . due to the very small thickness ( 5 nm ) of the lcmo layer in @xmath12 , the fm order of the mn moments is strongly suppressed . the dc magnetization data yield a very small moment of about 0.2@xmath7/mn . to confirm that the suppression of the fm order is due to the reduced lcmo layer thickness , we have grown a bl with a 10 nm thick lcmo layer under identical conditions and found that it has a fm magnetization comparable to @xmath10 and @xmath11 , and a curie temperature of @xmath30@xmath33k . a similar threshold effect of the fm properties depending on the thickness of the manganite layers has been reported in ref . @xcite . the resistance and magneto - resistance were also measured with a ppms using a four probe method with the wires glued with silver paint to the corners of the sample surface . the current was set to 10@xmath34 while the voltage was recorded . the temperature was changed at a rate of 2k per minute , the magnetic field was varied from @xmath35 t to @xmath36 t at a rate of 100oe per second . figure [ fig - rt - rh ] ( a ) shows that the ybco layers of yl_1 - 3 have fairly sharp superconducting transition with the critical temperatures ( @xmath0 ) of 70 - 75k . on the other hand , the resistance data show that the conductivity of the lcmo layers in @xmath11 is significantly lower than the one in @xmath10 . this is evident from the absence of a kink feature in the r - t curve in figure [ fig - rt - rh ] ( a ) and the much lower magneto - resistance effect in the r - h curve in the vicinity of @xmath30 in figure [ fig - rt - rh ] ( b ) . ( a ) temperature dependence of the field cooled magnetization at @xmath37 t for mls yl_1 - 3 . the data below @xmath38k have been omitted since they are strongly affected by vortex pinning and related avalances effects that lead to a macroscopic inhomogeneity of the magnetization . instead we extrapolated the magnetization curves according to @xmath39 as shown by the dashed lines . ( b ) m - h loops for samples yl_1 - 3 at @xmath40k measured after field cooling at @xmath41 t ] ( a ) resistance versus temperature in zero magnetic field for yl_1 - 3 . ( b ) magneto - resistance , @xmath42 at @xmath43k for @xmath10 and @xmath11 . ] [ xas ] the x - ray absorption spectroscopy ( xas ) measurements have been performed at the xtreme beamline of swiss light source at the paul scherrer institut in switzerland . the absorption spectra have been recorded simultaneously in the total electron yield ( tey ) and the total fluorescence yield ( tfy ) modes . for the xmcd measurements at 2k , the samples have been field cooled in 6 t applied parallel to the incident x - ray beam at a 30@xmath13 incident angle with respect to the sample surface . for each xmcd spectrum , we have changed the polarization of the incoming x - rays ( left and right circular ) as well as the direction of the applied field , @xmath44 . multiple xmcd measurements have been carried out to check reproducibility and to enhance the signal to noise ratio . finally , the xmcd is defined as the difference between two absorptions for which the angular momentum of the incoming x - ray photons parallel ( @xmath45 ) and antiparallel ( @xmath46 ) to @xmath44 . the presented xmcd spectra are normalized with respect to the maximum of @xmath47 , and expressed in percentage . the xmcd field scans have been obtained by measuring at two specific energies near the maximum of the @xmath48-edge and off - resonance near the pre - edge , respectively . we have verified that the difference between these two values represents the background subtracted signal . the xmcd field scans are scaled with respect to the corresponding high field xmcd spectra . due to the remanence of the superconducting magnet , we could not obtain reliable data below @xmath49 t . the xld spectra were obtained at 2k and @xmath50 t using linearly polarized x - rays with the electric field vector along the vertical ( @xmath51 polarization ) and horizontal ( @xmath52 polarization ) direction with respect to the plane of incidence . the incidence angle correction was performed according to : @xmath53 ; @xmath54 ; xld @xmath55 . multiple sets of measurements were carried out to check reproducibility of the xld spectra . finally , the representative xld spectra have been normalized with respect to the maximum of @xmath56 and are expressed in percentage . [ pnr ] the polarized neutron reflectometry ( pnr ) experiment on the ml @xmath11 with 10 bl repetitions has been performed at the nrex beamline of the frmii reactor in munich , germany , using a monochromatic neutron beam with a wavelength of 4.28 . at 300k@xmath57@xmath30@xmath32k , where the sample is not yet ferromagnetic , the reflectivity curve has been measured in unpolarized mode . subsequently , the sample was cooled to 5k in a field 0.1 t field that was applied parallel to the sample surface . the reflectivity profiles at 5k have been obtained for the spin - up and spin - down states of the neutrons . the data have been fitted with the same model of block - like nuclear and magnetic potentials as in refs . @xcite using the superfit program from the max planck institute in stuttgart that is based on a @xcite . at first , we deduced the structural parameters from the fitting of the 300k data . these were later used as input parameters for the fitting of the spin - polarized 5k data from which the magnetic depth profile has been obtained . [ opt ] the optical conductivity of the mls @xmath10 and @xmath11 has been measured with broad - band spectroscopic ellipsometry . in the near - infrared to ultraviolet range ( 0.5 - 6.5ev ) , we used a commercial ellipsometer ( woollam vase ) equipped with an ultra - high vacuum liquid he - flow cryostat . in the far - infrared and mid - infrared , we used a home - built setup , @xcite attached to a bruker 13v ftir spectrometer with a glowbar source . the correction for the response of the substrate has been performed with the woollam vase software @xcite . in subsection [ suppression_af - ei ] we present the x - ray absorption data ( xmcd and xld ) on the mls yl_1 - 3 which reveal that the strength of the antiferromagnetic exchange interaction ( af - ei ) between the cu and mn moments can be strongly suppressed whereas the magnitude of the cu moments is hardly reduced . in subsection [ two - component ] we discuss the evidence that the major part of this cu moment originates from the interfacial cu ions . in the following , we consider different possibilities to explain the strong reduction of the af - ei between the interfacial cu and mn moments in @xmath11 as compared to @xmath10 . in subsection [ result - pnr ] we show , based on polarized neutron reflectometry data , that this reduction is not related to a corresponding suppression of the ferromagnetic order of the interfacial mn moments . in subsection [ result - optics ] we present the evidence , based on transport , optical and xld data , that the strength of the af - ei is rather related to a change of the electronic properties of the lcmo layers . in particular , that the orbital polarons in the poorly conducting lcmo layers of @xmath11 are strongly reducing the af - ei . finally , in subsection [ result - yltheory ] we provide an explanation of the intrinsic nature of the ferromagnetic cu moment in terms of the modification of the intra - planar cu - cu exchange interaction that is brought about by the so - called orbital reconstruction of the interfacial cu ions . figure [ fig - cu - xmcd ] summarizes the xas data of the samples yl_1 - 3 which reveal that the strength of the af - ei between the cu and mn moments can be strongly suppressed whereas the magnitude of the cu ions is hardly affected . figure [ fig - cu - xmcd ] ( a ) displays a sketch of a ybco / lcmo ml and the xmcd experiment which selectively probes the ordered moment of the cu or mn ions ( along the x - ray propagation vector ) . the xmcd signal , i.e. , the difference between the absorptions for the left and right circularly polarized x - rays , has been obtained in the total electron yield ( tey ) and total fluorescence yield ( tfy ) modes . due to the limited escape depth of the photo - electrons of only few nanometers , the tey mode is very surface sensitive . since the mls are terminated with lcmo , the tey cu xmcd signal is governed by the cu ions at the uppermost ybco / lcmo interface . the tfy mode has a much larger probe depth and thus is equally sensitive to the bulk - like cu ions away from the interface . figure [ fig - cu - xmcd ] ( b ) displays the xmcd at the cu @xmath58-edges in the tey mode for a low field of 0.5 t and high fields of 5 or 6 t . figure [ fig - cu - xmcd ] ( c ) shows the corresponding magnetic field scans for the xcmd signals at the cu and mn edges . they have been taken at the energies where the xas curves in figure [ fig - cu - xmcd ] ( b ) exhibit the maximal xmcd signal . the positive ( negative ) xmcd ( for the applied field , @xmath59 ) indicates a magnetic moment that is parallel ( antiparallel ) to @xmath44 . for @xmath10 , the cu xmcd is always negative and almost saturates above 1 t . the corresponding mn xmcd is positive with a similar saturation behavior . this is the signature of the antiparallel orientation of the cu and mn moments that was interpreted in terms of a strong af - ei between cu and mn @xcite . for @xmath11 , the cu xmcd exhibits a remarkably different behavior . it is also negative at first , but above 1 t it reveals a paramagnet - like trend with a zero crossing around 3 t . the sign change is also evident from the cu xmcd curves at 0.5 t and 6 t in figure [ fig - cu - xmcd ] ( b ) ( middle panel ) . note that these trends have been confirmed for three different @xmath11 type samples . the observed behavior is characteristic of a very weak af - ei between cu and mn that is eventually overcome by the zeeman - interaction due to @xmath44 . strikingly , similar xmcd field scans have indeed been reported for a molecular system with a weak af coupling between cr and dy moments @xcite . finally , for @xmath12 , the cu xmcd signal is always positive suggesting that the af - ei with mn is entirely suppressed . this suppression of the af - ei is expected since the mn moments themselves are hardly ferromagnetic . the surprising result is that there is still no sign of a reduction of the magnitude of the maximal cu xmcd signal and thus of the interfacial cu moment . for the three kinds of samples yl_1 - 3 , the sum rule @xcite analysis yields rather similar effective spin moments of @xmath60@xmath7 per interfacial cu ion . first of all , it is important to clarify whether both the negative and the positive cu xcmd signals originate from the cuo@xmath3 planes next to the interface with lcmo or whether the latter is due to the bulk - like cu ions in the more distant cuo@xmath3 planes . the second possibility that the paramagnetic cu xcmd originates from bulk - like cu ions , has been suggested in ref . @xcite which reported similar xmcd data for corresponding lsco / lcmo mls . nevertheless , as shown in the following , we can exclude such a two component scenario for the case of these ybco / lcmo mls . the evidence is obtained from the analysis of the multi - peak structure of the xas curves ( a detailed description can be found in appendix [ app - multipeak ] ) . as shown in figure [ fig - cu - xmcd ] ( d)-(g ) , the interfacial and bulk - like cu ions give rise to distinct peaks @xcite in the xas curves . in particular , the low - energy peak around 930.4ev ( for which the position is marked by a dotted line in figure [ fig - cu - xmcd ] ( d)-(g ) ) originates from the interfacial cu ions since it is much stronger in the tey mode ( figure [ fig - cu - xmcd ] ( e ) ) which primarily probes the cu ions at the topmost ybco / lcmo interface , than in the tfy mode ( figure [ fig - cu - xmcd ] ( d ) ) . the negative xld ( @xmath61 ) of the @xmath62 ev peak , as compared to the positive one of the peaks at 931.0 and 931.7ev due to the bulk - like cuo@xmath3 planes , is the fingerprint of the orbital reconstruction of the interfacial cuo@xmath3 plane @xcite . the red - shift of the peak at 930.4ev can be understood in terms of a charge transfer between lcmo and ybco @xcite . the cu xmcd curves of @xmath11 in figure [ fig - cu - xmcd ] ( f ) and [ fig - cu - xmcd ] ( g ) highlight that the xmcd originates predominantly from the peak at 930.4ev and thus from the interfacial cu ions . this applies to the positive cu xmcd signal at 6 t as much as to the negative one at 0.5 t . in figure [ fig - cu - xmcd ] ( g ) there is also a very weak paramagnetic contribution from the peak at 931ev due to the bulk - like cu ions . this paramagnetic signal is however at least an order of magnitude smaller than the one from the interfacial cu ions . note that this conclusion is independent of the particular fitting procedure that is used to account for the different peaks . the essential effect is even seen in the bare spectra where the position of the peak of @xmath63 in tey mode ( solid blue circles in figure [ fig - cu - xmcd ] ( e ) ) coincides with the one of the maximum in the xmcd signal ( solid gray symbols in figure [ fig - cu - xmcd ] ( f ) and ( g ) ) . an additional argument against the two - component scenario of ref . @xcite is that such a paramagnetic signal is not observed in @xmath10 , despite the circumstance that this sample contains the same amount of bulk - like cu ions as @xmath11 . this puts the focus on the magnetic properties of the interfacial cu ions and the questions , firstly , why the af - ei with the mn moments on the lcmo side is so strongly reduced in the @xmath11 type samples and , secondly , why the magnitude of the ferromagnetic cu moment is independent of the strength of the exchange coupling with mn . concerning the first question we note that , based on the x - ray diffraction and reflectometry data , the structural quality , and the roughness of the ybco / lcmo interfaces are comparable for the @xmath10 and @xmath11 ( 10 bl repetitions ) type samples . furthermore , we find that whereas the roughness of the respective interface of the @xmath11 type sample with 1 bl repetition is only about 6 ( discussed in figure [ fig - xrd ] ( b ) ) , the characteristics of the cu xmcd signal is very similar to the one shown ( in figure [ fig - cu - xmcd ] ) for the @xmath11 10 bls sample that has a larger roughness of 10 . therefore , the strong suppression of the af - ei in @xmath11 is not the result of an increased interface roughness . furthermore , the tem study in figure [ fig - tem - eels ] has shown that @xmath11 has the same interfacial stacking ( termination ) as it was reported in ref . @xcite for @xmath10 , i.e. , with a sequence of cuo@xmath3-bao - mno@xmath3 layers that results in a straight cu - o@xmath64-mn bond across the ybco / lcmo interface ( and vice versa for the lcmo / ybco interface ) . this kind of interface termination is corroborated by the cu xld data which show no significant differences between these samples ( see figure 1 ( a)-(d ) of ref . @xcite for @xmath10 , figure [ fig - cu - xmcd ] ( d ) and ( e ) for @xmath11 , and figure [ cu - xld - yl3 ] ( a ) and ( b ) for @xmath12 ) . they suggest that these samples have a very similar charge transfer and orbital reconstruction of the interfacial cu ions . in particular , the orbital reconstruction of the interfacial cu ions , which is determined by the covalent bonding between the cu and mn ions , should be strongly affected by a change of the interfacial layer stacking . it was previously discussed that a strong covalent bonding requires a direct cu - o@xmath64-mn bond @xcite and shown that other kinds of interfacial layer stacking result in a much weaker orbital reconstruction effect @xcite . next , one may suspect that in @xmath11 the fm order of the mn moments is more strongly suppressed at the interface with ybco than in @xmath10 . this may be a result of the different growth and annealing conditions of the lcmo layers which in @xmath11 may weaken the fm double exchange interaction between the mn moments and strengthen the competing af interactions . such a reduction of the mn moment in the vicinity of the interfaces with ybco has already been observed in the @xmath10 sample and was discussed in terms of a _ _ naturally , one might assume that this suppression of the fm order of the interfacial mn moments is even stronger in @xmath11 . however , the pnr study of the same @xmath11 sample for which the xmcd data are shown in fig [ fig - cu - xmcd ] reveals the opposite trend , i.e. , it shows that the thickness of the _ _ in @xmath11 is smaller than the one in @xmath10 . to enable a direct comparison , the pnr data of @xmath11 in figure [ fig - pnr - yl2 ] have been fitted using the same block - like profile of the magnetic potential that was used in ref . @xcite for @xmath10 . this simplistic model assumes a complete suppression of the fm order of the mn moment in the _ _ . therefore , it does not allow for the more realistic scenario of a gradual decrease of the fm moment and a finite value of the mn moment at the interface . a stronger suppression of the interfacial mn moment is thus accounted for in terms of a larger thickness of the _ _ , @xmath65 . the fitting shown in figure [ fig - pnr - yl2 ] ( a ) yields _ _ t@xmath66__@xmath67 nm in @xmath11 that needs to be compared to _ _ t@xmath66__@xmath68 nm in @xmath10 ( the value is taken at the upper ybco / lcmo interface @xcite that is probed in tey mode ) . this trend has been confirmed with pnr measurements for an additional set of @xmath10 and @xmath11 type samples . this shows that a suppression of the fm order of the mn moment near the interface can not explain the much weaker af - ei in @xmath11 . ( a ) polarized neutron reflectometry data ( symbols ) for @xmath11 in the non - magnetic state at @xmath69k and the fm state at @xmath70k . also shown are the best fits ( solid lines ) using a block - like depth profile of the nuclear and magnetic potentials . ( b ) the deduced nuclear ( orange ) and magnetic ( blue ) depth profiles in units of the scattering length density ( sld ) . the magnetic moment in the central part of lcmo amounts to about @xmath71@xmath7 per mn ion . ] this puts the emphasis on the exchange coupling mechanism across the interface . it was previously shown that the ybco / lcmo mls have an interface termination with a layer stacking sequence of cuo@xmath3-y - cuo@xmath3-bao - mno@xmath3-(la , ca)o at both the ybco / lcmo and lcmo / ybco interfaces ( one plane of cuo chains per ybco layer is missing ) @xcite . this termination thus yields a covalent cu - o - mn bond across the interface which according to ref . @xcite is at the heart of the charge transfer and the orbital reconstruction of the interfacial cu ions . the corresponding feature in the cu xas spectra is the red - shifted peak at 930.4ev that is strongly enhanced in tey mode and exhibits only a weak cu xld signal ( as opposed to the large cu xld signal of the bulk - like cu ions ) . we find that these characteristic features due to the charge transfer and the orbital reconstruction of the interfacial cu ions are equally present in the cu xld spectra of the samples @xmath10 ( see figure 1 ( a)-(d ) of ref . @xcite ) , @xmath11 ( see figure [ fig - cu - xmcd ] ( d ) and ( e ) ) and @xmath12 ( see appendix [ app - xld - yl3 ] ) . in return , this suggests that these samples do not exhibit any significant differences concerning the interface termination , the resulting charge transfer and the orbital reconstruction of the interfacial cu ions . in the following we show that the most significant changes occur indeed with respect to the electronic properties of the lcmo layers . a clear reduction of the conductivity of the lcmo layers in @xmath11 , as compared to the one in @xmath10 , is evident from electric transport data in figure [ fig - rt - rh ] and also from the infrared spectroscopy data in figure [ fig - optics ] . the r - t data in figure [ fig - rt - rh ] ( a ) show that the kink in the resistance around @xmath30 , which is a signature of the transition from a paramagnetic insulator to a ferromagnetic metal in the lcmo layers , is fairly pronounced for @xmath10 whereas it is hardly visible for @xmath11 . the r - h curves in figure [ fig - rt - rh ] ( b ) show that the corresponding magneto - resistance effect in the vicinity of @xmath30 is much smaller for @xmath11 than for @xmath10 . finally , the infrared spectra in figure [ fig - optics ] ( b ) confirm that the increase of the drude peak below @xmath30 , which is a hallmark of the concomitant insulator - to - metal and paramagnet - to - fm transition of lcmo @xcite , is much weaker in @xmath11 than in @xmath10 . for the former a significant fraction of the spectral weight is instead accumulated in a broad mid - infrared ( mir ) band . the development below @xmath30 of such a mir band is well known from bulk manganites that are in the less hole doped part of the phase diagram close to the insulating fm ( i - fm ) phase where the charge carriers start to form fm polarons @xcite . the signatures of this mir band are also seen in a corresponding ml with lamno@xmath72 ( lmo ) layers that are known to be in the i - fm state @xcite ( see figure [ fig - optics ] ( b ) ) . this mir band is therefore a fingerprint of the fm polarons which are predicted to give rise to a particular local orbital order @xcite which involves an alternating occupation of the @xmath73-@xmath74 orbitals of the mn@xmath75 ions as shown in figure [ fig7-polaron - coupling ] ( a ) . notably , the orbital occupation can have a profound effect on the exchange interaction with the interfacial cu ions . this is shown by the scheme in figure [ fig7-polaron - coupling ] ( b ) which illustrates that the exchange interaction is af if the mn @xmath4 level is occupied , whereas it becomes fm if the in - plane polarized orbital is occupied . shown , for simplicity , is the extreme case for which the orbital reconstruction yields a complete inversion of the occupation of the cu @xmath76 and @xmath4 orbitals . ( a ) real part of the optical conductivity spectra ( @xmath77 ) of @xmath10 , @xmath11 and a [ ybco(10nm)/lamno@xmath72(10nm)]@xmath78 multilayer for which lmo is in the insulating fm state @xcite . ( b ) corresponding difference spectra of the optical conductivity , @xmath79 . @xmath80 of ybco / lmo ml reveals a pronounced mir band with a maximum around @xmath81@xmath82 that is characteristic of the orbital polarons . the inset magnifies the low energy part of @xmath80 . ] the strong af - ei in @xmath10 accordingly can be understood in terms of a preferred occupation of the mn @xmath4 orbitals . such an effect is indeed suggested by the tey mn xld data in figure [ figure8-mnxld ] ( a ) which yield an electron polarization of the @xmath83-orbitals of p@xmath84@xmath85% ( as detailed in the appendix [ app - peg ] ) . on the other hand , the scheme in figure [ figure8-mnxld ] ( b ) shows how the presence of the fm polarons in @xmath11 can reduce the strength of the af - ei . it displays the spatial arrangement of the occupied cu and mn @xmath83-orbitals close to the interface due to a fm polaron lattice for the representative case of a doping of @xmath86 . it gives rise to an alternation of the in - plane and out - of - plane polarization of the occupied mn @xmath83-orbitals along the lateral direction and , accordingly ( see figure [ fig7-polaron - coupling ] ( b ) ) , to a sign change of the exchange interaction with the cu ions . needless to say that the scheme in figure [ figure8-mnxld ] ( b ) shows a simplified and qualitative picture of how the presence of the fm polarons leads to a reduction of the af - ei between the interfacial mn and cu moments , even in the presence of a strong fm order of the mn moments . in the lcmo layers of @xmath11 , these fm polarons may still be partially dynamic and strongly disordered . nevertheless , especially in the vicinity of the interfaces , they can be pinned and thus contribute significantly to the suppression of the net af - ei with the cu moments . the presence of these fm polarons can also explain the smaller thickness of the _ _ in @xmath11 since they make the fm order of the mn moments more robust against the interfacial strain and disorder effects . finally , we remark that in line with this orbital polaron scenario , for which the average polarization of the mn @xmath83-orbitals should vanish ( see appendix [ app - polaron - xld ] ) , the mn xld signal of @xmath11 in figure [ figure8-mnxld ] ( a ) yields a reduced value of p@xmath84@xmath87% . ( a ) the quantity @xmath88the number of holes in the orbitals @xmath89 ( per site)as a function of the parameter @xmath90 . ( b ) @xmath90 dependence of the total spin of the cluster . in both panels , the solid red line corresponds to the ground state and the green dashed line to the first excited state . ( c ) energies of the ground state and of the first excited state as functions of the parameter @xmath90 . ] this brings us to the second question , that has been formulated at the beginning of subsection [ result - pnr ] , about the origin of the intrinsic cu moment of the interfacial cuo@xmath3 plane . we suggest that this is due to its strongly underdoped state which arises from the electron transfer from lcmo to ybco and the lack of a cuo chain layer which serves as a charge reservoir @xcite and to the orbital reconstruction . in a bulk - like environment , where the holes reside mainly in cu @xmath76 orbitals , there would be a strongly af intra - planar exchange interaction between the cu moments which suppresses the corresponding cu xmcd signal . nevertheless , for the cu ions in the cuo@xmath3 plane next to the ybco / lcmo interface , the occupation of the @xmath83-orbitals is substantially different : the concentration of holes in the cu @xmath4 orbitals is approximately the same as in the cu @xmath76 orbtials . this is the so called orbital reconstruction effect that was reported in ref . @xcite and is also clearly evident from the xld curves in figures [ fig - cu - xmcd ] ( d ) and ( e ) . in the following we outline that this particular occupation of the cu @xmath73-@xmath83 orbitals weakens the intra - planar af interaction and may even induce a weak fm interaction . in order to obtain more insight into the origin of the magnetic moment and its relation to the orbital reconstruction , we have performed exact diagonalization calculations for a cluster containing four cu sites with one hole per site , described by the extended hubbard model @xcite . the hamiltonian in the hole representation reads @xmath91 @xmath92 @xmath93 + h.c.\biggr)+\ ] ] @xmath94 @xmath95 it involves cu @xmath76 orbitals ( operators @xmath96 , @xmath97 ) and cu @xmath4 orbitals ( @xmath98 , @xmath99 ) , the corresponding particle number operators are denoted by @xmath100 . the first line of eq . ( [ eq : hubbard ] ) contains the on - site one - hole terms . the symbols @xmath101 and @xmath90 denote the energies of the orbitals @xmath76 and @xmath4 , respectively . here , we set @xmath102 . the parameter @xmath90 controls the occupations of the orbitals : for high ( low ) values of @xmath90 the holes can be expected to reside mainly in the @xmath89 orbitals ( in the @xmath103 orbitals ) , for intermediate values comparable occupations can be expected . the second and the third lines of eq . ( [ eq : hubbard ] ) contain the hopping terms . the sums run over all pairs of nearest neighbours and @xmath104 is equal to 1 ( -1 ) for pairs oriented along the @xmath105 axis ( @xmath106 axis ) . finally , the fourth and the fifth lines of eq . ( [ eq : hubbard ] ) contain the on - site interaction terms . the third of them involves the hund s rule exchange and can be written as @xmath107 , \label{eq : hund}\ ] ] where @xmath108 , @xmath109 are the on - site spin operators and @xmath110 . the calculations have been performed using the open boundary conditions . the following values of the input parameters have been used : @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath116 . the dependence of the ground state properties on the remaining parameter @xmath90 has been investigated . figure [ fig9 - 10 ] ( a ) shows the expectation value @xmath88 of the number of holes in the @xmath89 orbitals ( per site ) as a function of @xmath90 . the solid red line corresponds to the ground state and the dashed green line to the first excited state . it can be seen , that for high ( low ) values of @xmath90 , @xmath88 of the ground state is close to 1 ( close to 0 ) , as expected . for intermediate values in the range from ca -0.2 ev to ca -0.05 ev , @xmath117 and @xmath118 . clearly , the ground state properties for this particular range may be of relevance in the context of considerations of orbitally reconstructed cuo@xmath119 planes at the ybco / lcmo interfaces . figure [ fig9 - 10 ] ( b ) shows the @xmath90 dependence of the total spin of the cluster in its ground state and in the first excited state . interestingly , with decreasing @xmath90 the spin state of the ground state changes from a singlet at high values of @xmath90 to a quintuplet ( spins of the four holes parallel ) to a singlet at low values of @xmath90 . the quintuplet occurs precisely in the same @xmath90-window as the 0.4-plateau in @xmath88 . for completeness we show in figure [ fig9 - 10 ] ( c ) the @xmath90 dependence of the energies of the ground state and of the first excited state . note the crossings between the singlet line and the quintuplet line . the trend of the total spin can be qualitatively understood as follows . for high values of @xmath90 the holes are located mainly in the @xmath89 orbitals and the standard superexchange mechanism involving virtual hoppings between these orbitals stabilizes the singlet ground state . in the large @xmath120 limit the ground state can be even described analytically @xcite . with decreasing @xmath90 , more and more holes enter the @xmath103 orbitals , and at certain critical value of @xmath90 , virtual hoppings within @xmath121 configurations begin to play a more important role than those within the @xmath122 configurations and stabilize the ferromagnetic ( quintuplet ) ground state . the reason for why the former hoppings support the ferromagnetic configuration is that for parallel spins the energies of the intermediate states are lower than for antiparallel spins because of the hund s rule coupling ( for a schematic representation of the mechanism , see figure [ s4c ] ) . the in the @xmath90 dependence of @xmath123 appears only for relatively high values of @xmath124 , for the present values of @xmath125 and @xmath120 , @xmath124 has to be larger than @xmath126 . for very low values of @xmath90 , the holes are located mainly in the @xmath103 orbitals and the superexchange mechanism involving virtual @xmath127 hoppings yields the singlet ground state . in conclusion , results of our calculations demonstrate that the orbital reconstruction can give rise to a weak intra - planar ferromagnetic interaction between the spins of the interfacial cu ions . note that the 3d versions of the @xmath74 hubbard model are known to exhibit ferromagnetic solutions for certain ranges of doping , in particular for 1/4 filling and 3/4 filling , see , e.g. , ref . we have shown , based on x - ray absorption spectroscopy and polarized neutron reflectometry measurements , that the antiferromagnetic exchange interaction ( af - ei ) between the interfacial cu and mn moments in ybco / lcmo multilayers can be strongly suppressed , whereas the ferromagnetic moment of the interfacial cu ions remains sizeable . this suggests that the cu moments are not induced by the af - ei with mn but are intrinsic to the interfacial cuo@xmath3 planes . we have outlined that a weakly ferromagnetic intra - planar magnetic exchange interaction between the cu moments may arise due to the nearly equal hole occupation of the cu @xmath4 and the @xmath76 orbitals that is brought about by the so - called orbital reconstruction which originates from the hybridization with the mn ions . we have furthermore suggested that the strong suppression of the af - ei between the cu and mn moments may be caused by ferromagnetic polarons which develop in poorly hole doped lcmo layers . in terms of applications , these findings may be used to create spin - active cuprate / manganite interfaces for which the relative orientation of the cu and mn moments can be readily varied with an external magnetic field . this kind of interfacial spin control can for example allow one to induce a spin - triplet superconducting order parameter which serves as source of spin - polarized supercurrents . the fabrication of mesoscopic devices in which these ideas can be tested remains a project for future research . the work at unifr has been supported by the schweizer nationalfonds ( snf ) through grants no . 200020 - 153660 and crsii2 - 154410/1 . the work at muni was supported by the projects ceitec(cz.1.05/1.1.00/02.0068 ) and muni / a/1496/2014 . the xas measurements have been performed at the epfl / psi xtreme beamline of the swiss light source at the paul scherrer institut , villigen , switzerland . the neutron experiment has been done at the nrex instrument operated by the max - planck society at the heinz maier - leibnitz zentrum ( mlz ) , garching , germany and was supported by the european commission under the @xmath128 framework programme through the research infrastructures " action of the capacities programme , nmi3-ii , grant agreement number @xmath129 . [ app - multipeak ] the method of the multi - peak fitting of the xas curves in tey and tfy modes near their @xmath130-edge is discussed here . we have followed the same approach as detailed in ref . @xcite . for the cu atoms , we consider four transitions . these are ( a ) and ( b ) the @xmath131 transitions of the interfacial and the bulk @xmath132 ions , respectively , and ( c ) and ( d ) the @xmath133 transitions related to the zhang - rice singlets of bulk cu ions in the @xmath134 planes and cuo chains , respectively . accordingly , we used four lorentzian functions for the fitting of the xas near the @xmath130-edges [ eq . ] . to account for the edge - jumps and the non - zero background under the xas curves , we added a combination of linear and sigmodal functions [ eq . ] . to find the peak positions , all tfy and tey xas curves of a given sample have been fitted simultaneously with the peak positions as common parameters . in a next step , the peak positions have been fixed and each pair of xas curves ( @xmath135 and @xmath136 ; @xmath137 and @xmath138 ) are fitted simultaneously in tfy and tey modes by considering the parameters for background and widths of the lorentzian profiles as common parameters . in this way , we determined the four peaks around @xmath62 , @xmath139 , @xmath140 and @xmath141ev . the peak around @xmath62ev arises from the interfacial cu ions since its weight is small in the tfy mode but very large in the tey mode . [ app - xld - yl3 ] similar to @xmath10 and @xmath11 , the signature of the orbital reconstruction of the interfacial cu ions have been observed in @xmath12 as shown in figures [ cu - xld - yl3 ] ( a ) and ( b ) . the measurement has been done at 2k in remanence , after field cooling at 5 t . the xas curves in figures [ cu - xld - yl3 ] ( c ) and ( d ) show the corresponding xmcd curves at 0.5 and 5 t fields which confirm that the major part of the xmcd signal arises from the low energy peak at 930.5ev . cu xas curves of @xmath12 with the linear polarization of the x - rays parallel and perpendicular to the cuo@xmath3 planes in the ( a ) tfy and ( b ) tey modes . cu xas curves of @xmath12 for circular polarizations and the corresponding xmcd signals at ( c ) 0.5 t and ( d ) 5 t . ] [ app - peg ] the electron orbital polarization , @xmath144 of the lcmo layers has been deduced following the procedure of refs . @xmath145 here , @xmath146 and @xmath147 are the number of electrons in the @xmath148 and @xmath149 orbitals , respectively . exemplary xas curves for linear polarizations are shown in figure [ mn - xld - xas ] . mn xld curves of @xmath11 in tey mode showing @xmath150 and @xmath151 and base line , @xmath152 which accounts for edge jumps . ] the positive sign of @xmath144 indicates that @xmath149 has a higher hole concentration , i.e. , the electrons preferably occupy the @xmath148 orbital . [ app - polaron - xld ] it is shown that the ferromagnetic ( fm ) polarons do not give rise to x - ray linear dichroism signal ( xld ) , i.e. , the elctron polarization of their @xmath83-orbitals amounts to zero . as shown in figure [ fig7-polaron - coupling ] ( a ) , a fm polaron consists of two @xmath4 , two @xmath154 and two @xmath155 orbitals . among them , the last two orbitals are in - plane orbitals . in the following , they are represented in the basis of @xmath156 and @xmath157 . + + @xmath76:@xmath156 + @xmath4:@xmath157 + @xmath154:@xmath158 + @xmath155:@xmath159 + + calculation of the expectation value of the @xmath156 and @xmath157 thus yields : + + number of @xmath76 orbitals @xmath160 . + number of @xmath161 orbitals @xmath162 . + + the @xmath163 and @xmath4 orbitals thus have the same occupation probability which means that the xld and the electron polarization of the @xmath83-orbitals amount to zero . 45ifxundefined [ 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 10.1038/nmat3223 [ * * , ( ) ] link:\doibase 10.1126/science.1131091 [ * * , ( ) ] link:\doibase 10.1126/science.1146006 [ * * , ( ) ] link:\doibase 10.1103/physrevb.67.214511 [ * * , ( ) ] link:\doibase 10.1103/physrevb.69.134428 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.94.057002 [ * * , ( ) ] link:\doibase 10.1103/physrevb.71.140509 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.140407 [ * * , ( ) ] link:\doibase 10.1038/nphys272 [ * * , ( ) ] link:\doibase 10.1038/nmat2383 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.197201 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.137005 [ * * , ( ) ] link:\doibase 10.1038/nphys2318 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.134520 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.180506 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.115105 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.224509 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.205135 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.174415 [ * * , ( ) ] link:\doibase 10.1126/science.1149338 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.256804 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.196802 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.054514 [ * * , ( ) ] link:\doibase 10.1107/s0021889807045086 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.ultramic.2006.04.016 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.060405 [ * * , ( ) ] http://stacks.iop.org/0953-8984/20/i=26/a=264010 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/0168583x86902910 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0304885314005605 [ * * , ( ) ] link:\doibase 10.1103/physrevb.60.16073 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0040609004000033 [ * * , ( ) ] j.a . woollam co. inc . , @noop ( ) link:\doibase 10.1039/c2sc00794k [ * * , ( ) ] link:\doibase 10.1103/physrevlett.68.1943 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.75.152 [ * * , ( ) ] http://dx.doi.org/10.1038/ncomms6626 [ * * ( ) ] , http://dx.doi.org/10.1038/ncomms3336 [ * * ( ) ] , link:\doibase 10.1103/physrevb.55.4206 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.174422 [ * * , ( ) ] link:\doibase 10.1103/physrevb.60.13458 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0038110103002053 [ * * , ( ) ] link:\doibase 10.1103/physrevb.28.327 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.214431 [ * * , ( ) ] link:\doibase 10.1103/physrevb.72.214431 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.035112 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/0368204895025375 [ * * , ( ) ] _ _ , http://doc.rero.ch/record/232627[ph.d . thesis ] , ( )	with x - ray absorption spectroscopy and polarized neutron reflectometry we studied how the magnetic proximity effect at the interface between the cuprate high-@xmath0 superconductor @xmath1 ( ybco ) and the ferromagnet @xmath2 ( lcmo ) is related to the electronic and magnetic properties of the lcmo layers . in particular , we explored how the magnitude of the ferromagnetic cu moment on the ybco side depends on the strength of the antiferromagnetic ( af ) exchange coupling with the mn moment on the lcmo side . we found that the cu moment remains sizeable if the af coupling with the mn moments is strongly reduced or even entirely suppressed . the ferromagnetic order of the cu moments thus seems to be intrinsic to the interfacial cuo@xmath3 planes and related to a weakly ferromagnetic intra - planar exchange interaction . the latter is discussed in terms of the partial occupation of the cu @xmath4 orbitals , which occurs in the context of the so - called orbital reconstruction of the interfacial cu ions .
the discovery of the higgs boson at the large hadron collider ( lhc ) @xcite completes not only the standard model ( sm ) of particle physics , but gives also new constraints on any extension of it . the most studied extension of the sm to date is supersymmetry ( susy ) , see ref . @xcite and references therein . however , the minimal - supersymmetric standard model ( mssm ) is facing increasing pressure over the last few years . one of the reasons is the measured higgs mass of 125 gev which is not trivial to accommodate in the mssm . the tree - level mass of the higgs has the strict upper limit @xmath0 , and one therefore needs large radiative corrections to push the mass to the required value . the two main possibilities within the mssm are either heavy stops , and/or a large mixing among them . the first option raises immediately the question about naturalness @xcite , while the second one can cause charge and colour breaking minima @xcite . + for these reasons the interest in non - minimal susy models has increased in the last years . a widely studied approach to soften the need for large radiative corrections to the higgs mass is to increase the mass already at tree - level . this is done for instance in different singlet extensions via @xmath1-terms @xcite . the other option is to use @xmath2-terms @xcite . + the simplest realisations of these ideas show the expected decoupling behaviour : as soon as the new sector responsible for the enhanced tree - level mass becomes heavy , the positive effect of the higgs mass shrinks . however , it has been shown that this decoupling can be circumvented at the price of introducing large soft - breaking masses . indeed , valid models with non - decoupling @xmath1- @xcite or @xmath2-terms @xcite have been successfully constructed . + we will consider in this letter the case of non - decoupling @xmath2-terms via a new abelian gauge group @xmath3 . so far , only the leading order effect on the higgs mass was studied in this scenario : the impact of the non - decoupling @xmath2-terms at tree - level in the case of vanishing gauge kinetic mixing was considered . we will show that ( i ) radiative corrections and ( ii ) the presence of gauge - kinetic mixing can give important corrections to this approximation . + this letter is organised as follows : in [ sec : gkm ] a short excursion into gauge kinetic mixing in supersymmetric models is given , before in [ sec : model ] the origin of both effects in a concrete model is described . the impact on the higgs mass is analysed numerically in [ sec : numerics ] , before we summarise in [ sec : summary ] . if more than one abelian gauge group is present , mixing terms between the field strength tensors of the groups are allowed by gauge and lorentz invariance @xcite : @xmath4 here , @xmath5 are the field strength tensors of two different abelian groups @xmath6 , @xmath7 and @xmath8 is in general an @xmath9 matrix if @xmath10 abelian groups are present . for practical applications it is often useful to bring the propagators of the vector bosons back into their canonical form by the redefinition @xcite @xmath11 the trade - off for this redefinition is a change in the covariant derivative : @xmath12 where @xmath13 is the original diagonal matrix of @xmath10 individual gauge couplings associated to the @xmath10 abelian gauge factors , and @xmath14 is the vector of the relevant @xmath15 charges . thus , the @xmath16 factor can be translated into a new set of @xmath17 `` effective '' gauge couplings whose combinations populate off - diagonal entries of a gauge - coupling matrix @xmath18 in supersymmetric models , gauge kinetic mixing appears not only for the vector bosons , but also for both , the @xmath2-terms and the gaugino soft - breaking terms . of particular importance in the following are the new @xmath2-terms which read for the abelian sector @xmath19 while the non - abelian @xmath2-terms remain unchanged . + for the special case of two abelian gauge groups , the most general form of @xmath20 reads @xmath21 in general , one has the freedom to perform a change in basis to bring @xmath20 into a particular form . the most commonly considered cases are the symmetric basis with @xmath22 and the triangle basis with @xmath23 , @xmath24 . the triangle basis has the advantage that the new scalars do not contribute to the electroweak vev , and the entire impact of gauge kinetic mixing is encoded in one new coupling @xmath25 . the relations between @xmath26 ( @xmath27 ) and the physical couplings called @xmath28 , @xmath29 , @xmath25 are given by @xcite @xmath30 in order to show the impact of loop corrections and gauge kinetic mixing on the higgs mass in the presence of non - decoupling @xmath2-terms , we take a model inspired by ref . @xcite as example charges of the new fields to allow for a seesaw type - i via the term @xmath31 . in this model , the particle content of the mssm is extended by a gauge singlet superfield @xmath32 , a superfield for the right - handed neutrino and two fields @xmath33,@xmath34 responsible for the breaking of @xmath3 . it has also been shown that this model , when extended in addition by three generations of vector - like leptons and quarks , provides an attractive explanation for the diphoton excess observed at 750 gev by atlas and cms @xcite . .charges of the superfields under @xmath3 . the superfields in addition to the mssm particle content are singlets under the sm gauge group . [ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] the @xmath3 charges of all superfields are summarised in [ tab : u1xmssmparticles ] , while the most general superpotential in agreement with all global and local symmetries is given by @xmath35 for simplicity , we will neglect in the following @xmath36 , @xmath37 , and @xmath8 as well as their soft - breaking terms . in addition , the soft - breaking term of @xmath38 can always be shifted away . the relevant soft - breaking terms in the scalar sector are @xmath39 with @xmath40 . + electroweak symmetry breaking and the breaking of @xmath3 are triggered by the following scalars receiving vacuum expectation values ( vevs ) @xmath41 in addition , a vev for the scalar singlet is induced in general @xmath42 we define @xmath43 , @xmath44 as well as @xmath45 , and @xmath46 . one can use the five tadpole equations @xmath47 to eliminate five free parameters . we are doing this in the following for @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath38 . thus , the remaining free parameters in the scalar sector are @xmath52 , @xmath53 , @xmath54 , @xmath36 , @xmath55 and the vevs . in all numerical studies we are going to fix @xmath56 tev , and @xmath57 . the neutral higgs mass matrix can be written in the basis @xmath58 as @xmath59 with the abbreviations @xmath60 and @xmath61 , @xmath62 . we have used the relations @xmath63 , @xmath64,@xmath65 , @xmath66 , and took the limit @xmath67 , @xmath68 . the lightest eigenstate of the @xmath69 sub - system is approximated by @xmath70 here , we made use of @xmath71 and assumed @xmath72 . in addition , we replaced @xmath73 as well as @xmath74 for clarity by the full expressions . we find the expected limits * @xmath75 : @xmath76 * @xmath77 : @xmath78 thus , we have the desired @xmath2-term enhancement for very large soft - terms , which is , however , modified by the presence of the off - diagonal gauge coupling @xmath79 . as long as @xmath79 is small , the impact of gauge kinetic mixing will be negligible . thus , as long as @xmath79 is treated as free parameter , it can be set to zero at a given scale . however , via renormalisation group effects ( rge ) it will be induced radiatively if the two abelian gauge groups are not orthogonal . orthogonality in this respect means that the anomalous dimension matrix defined by @xmath80 is diagonal . here , @xmath81 and @xmath82 run over all @xmath15 groups and the trace runs over all superfields charged under the corresponding group . in many commonly studied @xmath15 extension as @xmath83 , @xmath84 is not diagonal @xcite . even if the two abelian groups can be embedded in a singlet gut group , gauge kinetic mixing might appear for the remaining _ light _ degrees of freedom @xcite . calculating the anomalous dimension matrix for the model under consideration , one finds large off - diagonal entries @xmath85 where @xmath86 is a diagonal matrix containing the gut normalisation . since we do nt assume any gut embedding here , we use @xmath87 in the following . at the one - loop level one can calculate the value of @xmath25 at the susy scale as function of ( i ) the scale @xmath88 where @xmath79 is assumed to vanish and ( ii ) the value for @xmath29 at this scale . the running of the gauge coupling matrix @xmath20 at one - loop is given by @xcite @xmath89 one has to ensure to bring @xmath20 to the particular basis defined in eq . ( [ eq : basis ] ) after the running . the result is shown in [ fig : gkm ] . at 1 tev as function of the scale @xmath88 where gauge kinetic mixing vanishes and @xmath29 at the scale . ] one finds that @xmath25 is in general negative , and only for a very low cut - off @xmath88 is the new coupling @xmath79 is much smaller than @xmath29 . for @xmath90 gev and large @xmath29 , @xmath91 can already be as large as 0.1 , while for a gut motivated scenario with @xmath92 and @xmath93 , @xmath79 can even be as large as 0.40.5@xmath94 . are considered , it can be wrong to neglect gauge - kinetic mixing . although the impact on the higgs mass might not be important , crucial effects for instance concerning @xmath95 searches @xcite or dark matter properties @xcite could be missed . ] for completeness , we want to mention that the possibility of positive ratios @xmath96 exists : this can appear for very large , but negative values of @xmath29 at the scale @xmath88 . because of the non - vanishing charge of the higgs doublets under the new gauge group , many additional radiative contributions to the higgs mass arise already at one - loop . the feynman diagrams with possible contributions proportional to @xmath29 are depicted in [ fig : diagrams ] . since the right - sfermions carry also a @xmath97 charge , they give also non - vanishing diagrams at one - loop . however , it can be shown that the tadpole and self - energy corrections cancel exactly for them . proportional to @xmath98 . the scalar contributions are summarised by @xmath99 , but for the case where only pseudo - scalars ( @xmath100 ) can appear . @xmath101 label the generations in the loop . ] the loop corrected scalar masses are the eigenvalues of the radiatively corrected mass matrix @xmath102 given by @xmath103 where @xmath104 is the sum of tadpole- and self - energy contributions . the pole masses @xmath105 are then associated with the eigenvalues of @xmath106 . + for the calculation of the diagrams it is necessary to diagonalise the mass matrices of the particles in the loop , which have dimensions up to @xmath107 . thus , most contributions can only be calculated numerically . only for the charged higgs contribution we can find an analytical estimate for the impact on the sm - like higgs mass which is mainly given by the change in the ( 2,2)-element of @xmath108 : @xmath109 @xmath110 is the renormalisation scale and we used the limit @xmath111 . in order to give an impression of the importance of all other loop corrections , we group all diagrams into four sets : ( i ) diagrams involving only neutralinos ( @xmath112 ) , ( ii ) diagrams involving only neutral cp - even scalars ( @xmath113 ) , ( iii ) diagrams involving only charged higgs scalars ( @xmath114 ) , ( iv ) diagrams with cp - odd scalars , the new gauge boson and the corresponding ghost ( @xmath100 , @xmath95 , @xmath115 ) . one has to consider the fields in the last group together , in order to have a gauge invariant set of diagrams . in [ fig : pis ] we show the values of the corrections to the ( 2,2)-element of the higgs mass matrix as function of @xmath29 for the case of vanishing kinetic mixing , @xmath116 tev and @xmath117 tev . as function of @xmath29 for the diagrams involving only ( i ) neutralinos ( green , dot - dashed ) , ( ii ) cp - even scalars ( blue , dashed ) , ( iii ) charged higgs ( orange , dotted ) , ( iv ) cp - odd scalars , @xmath95 and its ghost ( red , full ) . ] one can see that the main loop corrections come from the charged and cp - even scalar sector . since both are negative , one can expect that the new loop corrections reduce the sm - like higgs mass , i.e. they might compensate to some extent the positive effect at tree - level . the corrections stemming from neutralinos are comparable small , while the ones involving the @xmath95 , its goldstone boson and its ghost are even smaller . there is also an indirect effect of the extended gauge sector on the mssm - like corrections to the higgs mass : the new force changes the relation between the measured sm parameters and the running @xmath118values which enter the loop calculations . the most important parameter in this respect is the top yukawa coupling @xmath119 . as it can be seen from [ fig : yt ] , the change in the running coupling @xmath120 is only very small when turning up @xmath29 . thus , the change in the loop corrections is only very tiny . at the susy scale @xmath121 tev as function of @xmath29 . the red , dashed line is for @xmath122 tev and the blue , full line for @xmath123 tev . ] we turn now to the discussion of the impact on the sm - like higgs mass of the two effects described in the last section . for this purpose , we make use of the spheno@xcite output of sarah@xcite . the sarahgenerated fortrancode provides a fully fledged spectrum generator for the model which enables a calculation of the entire mass spectrum at the one - loop level without any approximation : any loop contribution and the full momentum dependence is included . in principle , sphenoincludes also new two - loop corrections for the model under consideration @xcite . however , these are so far only available in the gaugeless limit , i.e. the corrections we are mainly interested in are not covered . + we show in [ fig : gxgtmh ] the shifts of the sm - like higgs mass at tree - level and at the one - loop level for @xmath124 tev in the @xmath125-plane compared to the case @xmath126 . one sees that the enhancement in the higgs mass when including loop effects and/or gauge - kinetic mixing is always smaller than the push at tree - level when neglecting these effects . assuming a fixed ratio @xmath127 , gauge kinetic mixing becomes more important for larger @xmath29 . in contrast the ( relative ) impact of the loop corrections has only a mild dependence on the considered value of @xmath29 . this can be seen in [ fig : rel_t_l ] where the shifts @xmath128 and @xmath129 defined by @xmath130 are shown as function of @xmath29 and for different combinations of @xmath131 and @xmath54 . as defined in [ eq : d1,eq : d2 ] as function of @xmath29 . here , we put @xmath132 and assumed @xmath133 tev [ red , dashed],@xmath134 tev [ orange , dot - dotted],@xmath135 tev [ blue , dotted],@xmath136 tev [ green , full ] . ] one finds , that the loop effects always reduce the overall enhancement of the higgs mass by 2530% . the radiative corrections are usually more important for small @xmath29 , but also depend to some extent on the hierarchy between @xmath131 and @xmath54 : for larger @xmath54 , which corresponds to a large enhancement at tree - level , the relative importance of the loops decreases faster with increasing @xmath29 . we have revisited here the possibility to push the tree - level mass of the sm - like higgs by non - decoupling @xmath2-terms . it has been shown that a pure study at tree - level can overestimate the positive effect on the higgs mass significantly . namely , new radiative corrections triggered by the extended gauge sector can reduce the higgs mass by several gev . in addition , it has been shown that gauge kinetic mixing , which is a natural effect if the new abelian gauge group is not orthogonal to @xmath137 , reduces the positive shift of the higgs mass at tree - level to some extent depending on the assumed cut - off scale @xmath88 . i thank toby opferkuch for proof - reading the manuscript . 10 collaboration , s. chatrchyan _ et . al . _ , _ observation of a new boson at a mass of 125 gev with the cms experiment at the lhc _ , _ phys.lett . _ * b716 * ( 2012 ) 3061 [ http://arxiv.org/abs/1207.7235[1207.7235 ] ] . collaboration , g. aad _ et . _ , _ observation of a new particle in the search for the standard model higgs boson with the atlas detector at the lhc _ , _ phys.lett . _ * b716 * ( 2012 ) 129 [ http://arxiv.org/abs/1207.7214[1207.7214 ] ] . h. p. nilles , _ supersymmetry , supergravity and particle physics _ , _ phys . * 110 * ( 1984 ) 1162 . s. p. martin , _ a supersymmetry primer _ , http://arxiv.org/abs/hep-ph/9709356[hep-ph/9709356 ] . high energy phys.18,1(1998 ) ] . b. batell , g. f. giudice and m. mccullough , _ natural heavy supersymmetry _ , _ jhep _ * 12 * ( 2015 ) 162 [ http://arxiv.org/abs/1509.00834[1509.00834 ] ] . j. camargo - molina , b. oleary , w. porod and f. staub , _ stability of the cmssm against sfermion vevs _ , _ jhep _ * 1312 * ( 2013 ) 103 [ http://arxiv.org/abs/1309.7212[1309.7212 ] ] . j. camargo - molina , b. garbrecht , b. oleary , w. porod and f. staub , _ constraining the natural mssm through tunneling to color - breaking vacua at zero and non - zero temperature _ , _ phys.lett . _ * b737 * ( 2014 ) 156161 [ http://arxiv.org/abs/1405.7376[1405.7376 ] ] . u. chattopadhyay and a. dey , _ exploring mssm for charge and color breaking and other constraints in the context of higgs@125 gev _ , _ jhep _ * 1411 * ( 2014 ) 161 [ http://arxiv.org/abs/1409.0611[1409.0611 ] ] . m. bastero - gil , c. hugonie , s. f. king , d. p. roy and s. vempati , _ does lep prefer the nmssm ? _ , _ phys . lett . _ * b489 * ( 2000 ) 359366 [ http://arxiv.org/abs/hep-ph/0006198[hep-ph/0006198 ] ] . r. dermisek and j. f. gunion , _ consistency of lep event excesses with an h > aa decay scenario and low - fine - tuning nmssm models _ , _ phys . rev . _ * d73 * ( 2006 ) 111701 [ http://arxiv.org/abs/hep-ph/0510322 [ hep - ph/0510322 ] ] . u. ellwanger , c. hugonie and a. m. teixeira , _ the next - to - minimal supersymmetric standard model _ , _ phys.rept . _ * 496 * ( 2010 ) 177 [ http://arxiv.org/abs/0910.1785[0910.1785 ] ] . u. ellwanger and c. hugonie , _ the upper bound on the lightest higgs mass in the nmssm revisited _ , _ mod.phys.lett . _ * a22 * ( 2007 ) 15811590 [ http://arxiv.org/abs/hep-ph/0612133[hep-ph/0612133 ] ] . a. delgado , c. kolda , j. p. olson and a. de la puente , _ solving the little hierarchy problem with a singlet and explicit @xmath53 terms _ , _ phys . * 105 * ( 2010 ) 091802 [ http://arxiv.org/abs/1005.1282 [ 1005.1282 ] ] . g. g. ross and k. schmidt - hoberg , _ the fine - tuning of the generalised nmssm _ , _ _ * b862 * ( 2012 ) 710719 [ http://arxiv.org/abs/1108.1284[1108.1284 ] ] . l. j. hall , d. pinner and j. t. ruderman , _ a natural susy higgs near 126 gev _ , _ jhep _ * 04 * ( 2012 ) 131 [ http://arxiv.org/abs/1112.2703[1112.2703 ] ] . e. ma , _ exceeding the mssm higgs mass bound in a special class of u(1 ) gauge models _ , _ phys.lett . _ * b705 * ( 2011 ) 320323 [ http://arxiv.org/abs/1108.4029[1108.4029 ] ] . y. zhang , h. an , x .- d . ji and r. n. mohapatra , _ light higgs mass bound in susy left - right models _ , _ phys.rev . _ * d78 * ( 2008 ) 011302 [ http://arxiv.org/abs/0804.0268[0804.0268 ] ] . m. hirsch , m. malinsky , w. porod , l. reichert and f. staub , _ hefty mssm - like light higgs in extended gauge models _ , _ jhep _ * 1202 * ( 2012 ) 084 [ http://arxiv.org/abs/1110.3037[1110.3037 ] ] . x. lu , h. murayama , j. t. ruderman and k. tobioka , _ a natural higgs mass in supersymmetry from nondecoupling effects _ , _ phys.rev.lett . _ * 112 * ( 2014 ) 191803 [ http://arxiv.org/abs/1308.0792[1308.0792 ] ] . a. kaminska , g. g. ross , k. schmidt - hoberg and f. staub , _ a precision study of the fine tuning in the diracnmssm _ , _ jhep _ * 1406 * ( 2014 ) 153 [ http://arxiv.org/abs/1401.1816[1401.1816 ] ] . r. ding , t. li , f. staub , c. tian and b. zhu , _ the supersymmetric standard models with a pseudo - dirac gluino from hybrid @xmath138 and @xmath139term supersymmetry breakings _ , http://arxiv.org/abs/1502.03614[1502.03614 ] . p. batra , a. delgado , d. e. kaplan and t. m. p. tait , _ the higgs mass bound in gauge extensions of the minimal supersymmetric standard model _ , _ jhep _ * 02 * ( 2004 ) 043 [ http://arxiv.org/abs/hep-ph/0309149[hep-ph/0309149 ] ] . a. maloney , a. pierce and j. g. wacker , _ d - terms , unification , and the higgs mass _ , _ jhep _ * 06 * ( 2006 ) 034 [ http://arxiv.org/abs/hep-ph/0409127[hep-ph/0409127 ] ] . k. babu , i. gogoladze and c. kolda , _ perturbative unification and higgs boson mass bounds _ , http://arxiv.org/abs/hep-ph/0410085 [ hep - ph/0410085 ] . r. martinez , n. poveda and j. a. rodriguez , _ upper bound of the lightest higgs boson in a supersymmetric su(3)l x u(1)x gauge model _ _ * d69 * ( 2004 ) 075013 . e. bertuzzo and c. frugiuele , _ natural sm - like 126 gev higgs boson via nondecoupling d terms _ , _ phys . rev . _ * d93 * ( 2016 ) , no . 3 035019 [ http://arxiv.org/abs/1412.2765[1412.2765 ] ] . r. m. capdevilla , a. delgado and a. martin , _ light stops in a minimal u(1)x extension of the mssm _ , _ phys . rev . _ * d92 * ( 2015 ) , no . 11 115020 [ http://arxiv.org/abs/1509.02472[1509.02472 ] ] . b. holdom , _ two u(1 ) s and epsilon charge shifts _ , _ phys . _ * b166 * ( 1986 ) 196 . r. m. fonseca , m. malinsk and f. staub , _ renormalization group equations and matching in a general quantum field theory with kinetic mixing _ * b726 * ( 2013 ) 882886 [ http://arxiv.org/abs/1308.1674[1308.1674 ] ] . b. oleary , w. porod and f. staub , _ mass spectrum of the minimal susy b - l model _ , _ jhep _ * 05 * ( 2012 ) 042 [ http://arxiv.org/abs/1112.4600[1112.4600 ] ] . f. staub _ et . _ , _ precision tools and models to narrow in on the 750 gev diphoton resonance _ , http://arxiv.org/abs/1602.05581 [ 1602.05581 ] . l. basso , s. moretti and g. m. pruna , _ a renormalisation group equation study of the scalar sector of the minimal b - l extension of the standard model _ , * d82 * ( 2010 ) 055018 [ http://arxiv.org/abs/1004.3039[1004.3039 ] ] . m. hirsch , w. porod , l. reichert and f. staub , _ phenomenology of the minimal supersymmetric @xmath140 extension of the standard model _ , _ phys . rev . _ * d86 * ( 2012 ) 093018 [ http://arxiv.org/abs/1206.3516[1206.3516 ] ] . m. e. krauss , w. porod and f. staub , _ so(10 ) inspired gauge - mediated supersymmetry breaking _ , _ phys.rev . _ * d88 * ( 2013 ) , no . 1 015014 [ http://arxiv.org/abs/1304.0769[1304.0769 ] ] . p. athron , d. harries , r. nevzorov and a. g. williams , _ @xmath141 inspired susy benchmarks , dark matter relic density and a 125 gev higgs _ , http://arxiv.org/abs/1512.07040[1512.07040 ] . f. del aguila , g. d. coughlan and m. quiros , _ gauge coupling renormalization with several u(1 ) factors _ , _ nucl . _ * b307 * ( 1988 ) 633 . [ erratum : nucl . phys.b312,751(1989 ) ] . r. m. fonseca , m. malinsky , w. porod and f. staub , _ running soft parameters in susy models with multiple u(1 ) gauge factors _ , _ nucl.phys . _ * b854 * ( 2012 ) 2853 [ http://arxiv.org/abs/1107.2670[1107.2670 ] ] . l. basso , a. belyaev , s. moretti , g. m. pruna and c. h. shepherd - themistocleous , _ @xmath95 discovery potential at the lhc in the minimal @xmath142 extension of the standard model _ , _ eur . c71 * ( 2011 ) 1613 [ http://arxiv.org/abs/1002.3586[1002.3586 ] ] . m. e. krauss , b. oleary , w. porod and f. staub , _ implications of gauge kinetic mixing on z and slepton production at the lhc _ , _ * d86 * ( 2012 ) 055017 [ http://arxiv.org/abs/1206.3513 [ 1206.3513 ] ] . l. basso , b. oleary , w. porod and f. staub , _ dark matter scenarios in the minimal susy b - l model _ , _ jhep _ * 09 * ( 2012 ) 054 [ http://arxiv.org/abs/1207.0507[1207.0507 ] ] . w. porod and f. staub , _ spheno 3.1 : extensions including flavour , cp - phases and models beyond the mssm _ , _ comput . * 183 * ( 2012 ) 24582469 [ http://arxiv.org/abs/1104.1573 [ 1104.1573 ] ] . w. porod , _ spheno , a program for calculating supersymmetric spectra , susy particle decays and susy particle production at e+ e- colliders _ , _ comput . * 153 * ( 2003 ) 275315 [ http://arxiv.org/abs/hep-ph/0301101[hep-ph/0301101 ] ] . f. staub , _ sarah _ , http://arxiv.org/abs/0806.0538 [ 0806.0538 ] . f. staub , _ from superpotential to model files for feynarts and calchep / comphep _ , _ comput.phys.commun . _ * 181 * ( 2010 ) 10771086 [ http://arxiv.org/abs/0909.2863[0909.2863 ] ] . f. staub , _ automatic calculation of supersymmetric renormalization group equations and self energies _ , _ comput.phys.commun . _ * 182 * ( 2011 ) 808833 [ http://arxiv.org/abs/1002.0840[1002.0840 ] ] . f. staub , _ sarah 3.2 : dirac gauginos , ufo output , and more _ , _ comput.phys.commun . _ * 184 * ( 2013 ) pp . 17921809 [ http://arxiv.org/abs/1207.0906[1207.0906 ] ] . f. staub , _ sarah 4 : a tool for ( not only susy ) model builders _ , _ comput.phys.commun . _ * 185 * ( 2014 ) 17731790 [ http://arxiv.org/abs/1309.7223[1309.7223 ] ] . f. staub , _ exploring new models in all detail with sarah _ , http://arxiv.org/abs/1503.04200[1503.04200 ] . m. d. goodsell , k. nickel and f. staub , _ two - loop higgs mass calculations in supersymmetric models beyond the mssm with sarah and spheno _ , _ eur.phys.j . _ * c75 * ( 2015 ) , no . 1 32 [ http://arxiv.org/abs/1411.0675[1411.0675 ] ] . m. goodsell , k. nickel and f. staub , _ two - loop higgs mass calculation from a diagrammatic approach _ , http://arxiv.org/abs/1503.03098 [ 1503.03098 ] .	non - decoupling d - terms are an attractive possibility to enhance the tree - level mass of the standard model like higgs boson in supersymmetric models . we discuss here for the case of a new abelian gauge group two effects usually neglected in literature : ( i ) the size of the additional radiative corrections to the higgs mass due to the presence of the new gauge coupling , and ( ii ) the impact of gauge kinetic mixing . it is shown that both effects reduce to some extent the positive effect of the non - decoupling d - terms on the higgs mass . cern - th-2016 - 061
neutrino oscillations between muon and tau neutrinos , which were found recently in the super - kamiokande experiment @xcite , prove the existence of neutrino masses . this finding does not spoil the standard big bang nucleosynthesis ( bbn ) picture because experimentally interesting neutrino masses for all three flavors @xmath8 , @xmath9 , are small @xmath10 ev , in comparison to typical bbn temperatures @xmath11 mev . as for oscillations between active flavors , these do not play any role in bbn because all active flavors are equally populated . however , if sterile neutrinos exist , oscillations between active and sterile states could have significant consequences for bbn . one possibility is that the total energy density of the universe is changed due to excited sterile neutrinos ; together with a recent analysis of observational data @xcite , which claims that the effective number of neutrinos can not exceed @xmath12 , this effect can be used to exclude some region of the @xmath13 parameter plane e.g. for heavy sterile neutrinos @xcite . another possible influence of sterile neutrinos on bbn arises from the fact that @xmath14-@xmath15 oscillations can create a large asymmetry between @xmath14 and @xmath16 which would have a direct impact on the @xmath2-@xmath3 reactions and thus change the light element abundances . this case is interesting for small negative @xmath17 and small @xmath18 , and the important result will be the value of the electron neutrino asymmetry at the freeze - out of the @xmath2/@xmath3 ratio at a temperature around 1 mev . actually , the magnitude of @xmath14 asymmetry at @xmath19 is just an indicator for possible consequences on bbn . for accurate calculations one would need to use exact non - equilibrium distribution functions of @xmath14 and @xmath16 in the neutron - proton reactions . a general discussion of this subject can be found in section 2.1 of @xcite , and details in case of large asymmetry in @xcite . in early works with simplified calculations it was found that the asymmetry could not reach large final values @xcite . however , later it was found that the asymmetry can increase significantly and reach values @xmath20 @xcite . this finding was confirmed by elaborate numerical calculations @xcite . a main feature of these results was that the asymmetry grows according to the power law @xmath21 . a contradictory result that the power law was @xmath22 was suggested in @xcite , but was found to be erroneous in @xcite . in the present paper we investigate the evolution of the lepton asymmetry under the assumption that all active neutrinos passing the resonance are converted to sterile neutrinos , thus forming a fermi - dirac distribution in the sterile sector . this assumption is correct in the case of pure mikheyev - smirnov - wolfenstein ( msw ) transition @xcite , i.e. when one can neglect the collision - integral terms . although the assumption of full conversion from active to sterile states may be too strong for certain models , our result does present the maximum possible asymmetry . the advantage of our approach is that we do not solve any differential equations . thus , our approach is extremely simple and safe . furthermore , we consider the effects of various simplifications and compare our results with previous works , in particular with @xcite . finally , we investigate how the maximum amplitude of the asymmetry at the temperature 1 mev depends on initial conditions and @xmath5 , and compare it with a more detailed calculation , @xcite . we are considering neutrino oscillations in the early universe at temperatures around @xmath23 . in this epoch , the universe is a homogeneous and isotropic plasma , and the expansion is described by the hubble parameter , h = , where @xmath24 is the planck mass . the total energy density is _ tot = g_t^4 , where @xmath25 is the number of effective degrees of freedom during the epoch of interest . we also suppose that @xmath26 , which is a good approximation for the models we are considering here . the neutrino oscillations are described by the density matrix formalism , see for instance @xcite . however , we will mainly be interested in the resonance condition , which for @xmath27 oscillations is given by @xcite = - v_eff^a = - ( c_1 g_f t^3 + c_2^a ) , [ rescond ] where @xmath28 specifies the type of active neutrino , @xmath29 is the mass squared difference , @xmath30 is the mixing angle , and @xmath31 is the effective potential . the `` @xmath32 '' signs refer to neutrinos and anti - neutrinos , respectively . furthermore , @xmath33 gev@xmath34 is the fermi coupling constant , @xmath35 is the fine structure constant , @xmath36 , @xmath37 for @xmath38-@xmath15 mixing , @xmath39 for @xmath40-@xmath15 and @xmath41 is the temperature of the plasma . sometimes we will use the notation @xmath42 . finally , the effective asymmetry is @xmath43 , where @xmath44 is the neutrino asymmetry of the active species a , and @xmath45 is some small fermion asymmetry from all other fermion species . for our analysis it is only interesting that @xmath45 is of order of the baryon asymmetry , @xmath46 . because the temperature is much larger then the neutrino masses , not only neutrino momentum and energy are equal , @xmath47 , but also that a certain neutrino mode evolves with constant @xmath48 . we will therefore use this dimensionless variable instead of momentum and energy , so @xmath49 . we briefly discuss the evolution of the neutrino densities for small mixing angles , @xmath50 , looking at @xmath14 for definiteness . let us first sketch the reason why the lepton asymmetry can change at all . due to the mixing of @xmath14 with @xmath51 , the number densities of the @xmath14 and @xmath52 can alter . the sterile neutrinos that are thus created have no effect on @xmath1 . now if @xmath53 initially , the resonance condition for the @xmath14 and @xmath16 will be different . therefore their number densities will change differently , thereby changing @xmath1 . thus neutrino mixing works as a back - reaction on @xmath1 . the actual change in asymmetry and density takes place close to the resonance momentum , which is defined by eq . ( [ rescond ] ) , y_res= k_1 , [ yres ] where k_1= k_2= . there are three different stages of lepton asymmetry evolution . during the first stage the asymmetry is so small that the @xmath54 terms can be neglected in eq . ( [ yres ] ) . the neutrinos and the anti - neutrinos simultaneously pass through the resonance . furthermore , any initial asymmetry decreases due to the negative back - reaction . the second stage is the stage of exponential growth of the lepton asymmetry , which occurs when the back - reaction becomes positive . here , the resonance conditions for @xmath55 and @xmath56 are rapidly driven apart . for definiteness , we will assume that the created asymmetry is positive , @xmath57 . then @xmath58 for the @xmath55 increases to very high momenta where the fermi distribution is negligible , the resonance momentum for @xmath56 decreases to small momenta @xmath59 . at the same time , the asymmetry grows many orders of magnitude and reaches values of order of @xmath60 . in the third stage the lepton asymmetry grows at a slower rate . there has been a lot of controversy on this subject , and it has often be proposed that the lepton asymmetry would grow according to an approximate power law , @xmath61 , where @xmath62-@xmath63 . we will now show that any power law is only an approximation to the true asymmetry growth . our considerations start at the end of the exponential regime , which happens at temperature @xmath64 . the important variables are the asymmetry @xmath65 and the resonance momentum @xmath66 at this time . for definiteness , we will assume @xmath67 , so that we can safely neglect @xmath55 oscillations ( with @xmath68 we could neglect @xmath56 ) . the resonance momentum is governed by two effects : it decreases for increasing @xmath1 and increases for decreasing @xmath41 . at the end of the exponential regime , these two effects compensate each other , so that at first the resonance momentum changes very slowly . this plus the fact that collisional damping is small may let us assume that the momentum modes passing the resonance in the regime of slow asymmetry growth will fully excite the sterile sector . for negligible collisional damping , this is exactly the msw effect @xcite . the full excitation of the sterile sector at resonance has also been stated in @xcite . we can now calculate the asymmetry for a given temperature @xmath69 : ( t)=_i + _ y_i^y_res(t)dy y^2 f_eq(y , ) . [ resint ] here , @xmath70 is the equilibrium fermi distribution function , given by f_eq(y,)= , [ feq ] where @xmath71 is the dimensionless chemical potential for the active neutrinos . for @xmath72 and large temperatures @xmath73 , the chemical potential is approximately @xmath74 . therefore , we can safely neglect it . we will discuss its influence for large @xmath1 later on . we should mention that @xmath75 has its origin in an already weakly excited sterile sector at momenta @xmath76 . however , the correction to eq . ( [ resint ] ) from this excitation is negligible . we can now insert eq . ( [ yres ] ) into eq . ( [ resint ] ) and solve the equation for @xmath77 . the resulting equation only depends on the input parameters @xmath75 and @xmath78 , and on the neutrino parameter @xmath5 . interestingly , the equation is independent of the mixing angle @xmath79 , a fact which stems from the full transition assumption . the result has another striking feature : even if the sterile sector is not fully excited during resonance , which happens if the resonance momentum changes too fast and/or the collisional damping is important , our result still gives the maximally possible asymmetry at temperature @xmath41 . for small momenta , the fermi - dirac distribution can be approximated by @xmath80 . then eq . ( [ resint ] ) simplifies to ( t)-_i = (y_res^3(t)-y_i^3 ) . furthermore , for large asymmetries and small temperatures , the @xmath81 term in eq . ( [ yres ] ) is much smaller than the @xmath54 term , so that y_res . [ late_res ] [ yresapp ] then ^4 + ^3 ( y_i^3- _ i ) = ()^3 . [ eta4 ] this equation can be easily solved in two limits : when @xmath82 we can neglect the first term on the lhs so that we obtain a @xmath83 power law . however , when @xmath84 the solution is @xmath85 . thus the power law will slowly change in time . this is also what we expect from the following considerations : the asymmetry is created close to the resonance momentum . however , when the sterile sector is fully excited at this momentum , the asymmetry can no longer be created . naturally , when @xmath1 does not change , the resonance momentum will increase due to the decreasing temperature . thus we expect a slowly increasing resonance momentum . from eq . ( [ late_res ] ) we can then deduce that @xmath1 increases more slowly than a @xmath83 power law . we can easily derive an equation similar to eq . ( 14 ) of @xcite by differentiating eq . ( [ resint ] ) in the temperature @xmath41 : = f_eq(y_res,0 ) . [ detadt ] the only difference between our eq . ( [ detadt ] ) and eq . ( 14 ) of @xcite is in the usually negligible term proportional to @xmath86 . furthermore , eq . ( 18 ) of @xcite , which can be derived from their eq . ( 14 ) directly , is in our notation = , where we have used the simplification eq . ( [ late_res ] ) . ( 121,80 ) ( 20,0 ) ( 27,66)@xmath87 ( 130,4)@xmath88 ( 95,46)@xmath89 ( 96,30)@xmath90 ( 77,20)@xmath91 ( 55,20)@xmath92 ( 36,15)@xmath93 ( 121,80 ) ( 20,0 ) ( 27,66)@xmath87 ( 130,15)@xmath41 ( 95,50)@xmath89 ( 96,35)@xmath90 ( 79,28)@xmath91 ( 60,26)@xmath92 ( 38,24)@xmath93 for illustration , we have plotted the evolution of the `` power '' @xmath87 which can be derived by the formula = in fig . [ alpha_1 ] . here , we have already included the effect from the exact fermi - dirac distribution , which becomes important at low temperatures . we have plotted the evolution for @xmath94 and for different values of @xmath78 . note that typical values are @xmath95-@xmath92 . furthermore , we have set @xmath96 . typical values would be @xmath97-@xmath98 , but as we can see from eq . ( [ eta4 ] ) , such values are negligible in comparison with the @xmath99 term . anyhow , we can simulate the effect of @xmath75 by changing @xmath78 , see eq . ( [ eta4 ] ) . interestingly , the effect from @xmath5 can be compensated exactly by rescaling the temperature : t _ = ( ) ^1/4 t_=1 ^2 . to be still more precise , we need to use the exact eq . ( [ yres ] ) instead of eq . ( [ late_res ] ) . we have plotted the results in fig . [ alpha_2 ] in analogy to fig . [ alpha_1 ] . we see that both the approximate and the exact resonance condition give the same results at low temperatures , while at high temperatures the results are very different ; the exact result shows a larger power - law behavior . this means that there is no real power - law solution in the msw dominated region , at least under the assumption that transition is complete . we should mention that for @xmath40-@xmath15 oscillations , the correction due to the exact resonance condition is somewhat weaker , since @xmath100 is smaller by a factor 4 . we should note that using eq . ( [ late_res ] ) instead of eq . ( [ yres ] ) has no influence on the later evolution of the asymmetry , i.e. eq . ( [ late_res ] ) is a good approximation . this is because the apparently larger increase of @xmath1 for given @xmath64 , @xmath75 is compensated by a smaller @xmath78 when using eq . ( [ yres ] ) . in fig . [ alpha_2 ] we also show the effect of a non - zero initial asymmetry . this effect is important only for small initial values of the resonance momentum @xmath101 . ( 121,80 ) ( 20,0 ) ( 81,0)@xmath103 ( 128,58)@xmath104 ( 121,80 ) ( 20,0 ) ( 84,70)@xmath58 ( 128,2)@xmath104 if we want to investigate the influence of the neutrino asymmetry on bbn , we need to check its value at the temperature @xmath19 when the neutron - proton reactions decouple and the @xmath102 ratio freezes out . in fig . [ fnal_1 ] we present the dependence of the maximum possible asymmetry at the temperature 1 mev from @xmath104 . the solid line shows the maximum asymmetry for the initial conditions @xmath89 and @xmath96 . our result is in good agreement with previous extensive calculations @xcite . naturally , we do not expect any dependence on @xmath105 , since we have assumed full msw transition . we should therefore stress that our result is only a good approximation in regions of the parameter space ( @xmath5,@xmath105 ) where there is an effective exponential increase of the asymmetry , and if there is full msw transition afterwards . from fig . [ fnal_1 ] we can easily extract that for @xmath106 the asymmetry will remain small , while for @xmath107 it can reach its asymptotic value @xmath108 , which was found in @xcite ( the factor 2 difference with our @xmath109 on fig . [ fnal_1 ] stems from the relation @xmath110 ) . for @xmath111 , the asymmetry at temperature 1 mev can be fitted according to the simple relation = 0.1\|\|^3/4 in good agreement with the result of @xcite , @xmath112 shows the effect of a non - zero initial asymmetry ( for @xmath113 in this particular case ) . a non - zero initial asymmetry will increase the asymmetry at @xmath114 for small @xmath115 . a non - zero momentum @xmath78 will work in the opposite direction and decrease the final asymmetry for small @xmath104 ( long dashed line is for @xmath116 ) . note that for very small @xmath117 the exponential regime is to be expected at @xmath4 so that our results become invalid . finally , the chemical potential plays an important role when the asymmetry is large . however , its effect is to exponentially damp the number density of the @xmath52 ( in the case @xmath57 ) , thus decreasing the asymmetry . therefore , our result remains the maximum possible asymmetry . we have estimated the decrease of the asymmetry due to the chemical potential to be less than a factor of 2 . for illustration , fig . [ fnal_2 ] shows the dependence of the dimensionless resonance momentum @xmath118 at the temperature @xmath119 mev from the parameter @xmath104 . we see that at @xmath106 the resonance momentum is small , while for @xmath120 it is already on the tail of the distribution function . this explains the values of the asymmetry presented in fig . [ fnal_1 ] . in this work we have found the maximum lepton asymmetry which can be created at a given temperature due to active - sterile neutrino oscillations in the early universe . we have assumed full transitions of active to sterile neutrinos due to msw resonance to derive a very simple relation between the lepton asymmetry and the temperature . from this relation , we found that the evolution of the lepton asymmetry does not evolve according to a specific power law . we have also investigated on the maximum asymmetry at the temperature 1 mev when the @xmath102 ratio freezes out . we found that the value of the asymmetry at this time depends on @xmath5 , but not on @xmath121 in concordance with @xcite . furthermore , the initial settings described by the initial resonance momentum @xmath78 and asymmetry @xmath75 have only an influence for very small @xmath5 . this means that details of the previous evolution are not very important for the value of the asymmetry at @xmath122 . in the parameter region which is related to a large final asymmetry , @xmath123 , our asymmetry values are in good agreement with those of @xcite . we stress again that our results depend on the assumption of full transitions of active to sterile neutrinos . if in some part of the parameter space the transition is not total , then the asymmetry can be somewhat smaller . also , for large asymmetries , the exact value of @xmath1 may be smaller due to effects from the chemical potential . we are grateful to g. raffelt for helpful comments and reading the manuscript . we also would like to thank s. hansen , r. foot and r.r . volkas for useful comments . 99 sk super - kamiokande collaboration , y. fukuda _ et al . _ , ; ; ; . tytler0001318 d. tytler , j. m. omeara , n. suzuki and d. lubin , to appear in physica scripta , astro - ph/0001318 . dhrs a.d . dolgov , s.h . hansen , g. raffelt and d.v . semikoz , hep - ph/0008138 . karmen00 a.d . dolgov , s.h . hansen , g. raffelt and d.v . semikoz , . foot00 r. foot , . enkvist90 k. enqvist , k. kainulainen and j. maalampi , . barbieri91 r. barbieri and a. dolgov , . foot96 r. foot , m. thomson and r.r . fv96 r. foot and r.r . volkas , ; erratum . r. foot , . bfoot00 p. di bari and r. foot , . dolgov99 a.d . dolgov , s.h . hansen , s. pastor and d.v . semikoz , . we2000 r. buras , d.v . semikoz , hep - ph/0008263 comment00 p. di bari , r. foot , r. r. volkas , y. y. y. wong , hep - ph/0008245 l. wolfenstein , ; ; + s.p . mikheyev and a. yu . smirnov , . x. shi , g.m . fuller , k. abazajian , . dolgov81 a.d . notzold88 d. ntzold and g. raffelt , .	a large lepton asymmetry could be generated in the early universe by oscillations of active to sterile neutrinos with a small mixing angle @xmath0 . the final order of magnitude of the lepton asymmetry @xmath1 is mainly determined by its growth in the last stage of evolution when the msw resonance dominates the kinetic equations . in this paper we present a simple way of calculating the maximum possible lepton asymmetry which can be created . our results are in good agreement to previous calculations . furthermore , we find that the growth of asymmetry does not obey any particular power law . we find that the maximum possible asymmetry at the freeze - out of the @xmath2/@xmath3 ratio at @xmath4 strongly depends on the mass - squared difference @xmath5 : the asymmetry is negligible for @xmath6 and reaches asymptotically large values for @xmath7 . = -1.0 cm # 1#2 mpi pht/2000 - 37 + .06 in * maximum lepton asymmetry from active sterile neutrino oscillations in the early universe . * + * r. buras * + _ max - planck - institut fr physik ( werner - heisenberg - institut ) + fhringer ring 6 , 80805 mnchen , germany _ + * d.v . semikoz * + _ max - planck - institut fr physik ( werner - heisenberg - institut ) + fhringer ring 6 , 80805 mnchen , germany + and + institute of nuclear research of the russian academy of sciences + 60th october anniversary prospect 7a , moscow 117312 , russia _ +
directed polymers in a quenched random potential have been the subject of intense investigations during the past two decades ( see e.g. @xcite ) . in the one - dimensional case we deal with an elastic string directed along the @xmath1-axis within an interval @xmath2 $ ] . randomness enters the problem through a disorder potential @xmath3 $ ] , which competes against the elastic energy . the problem is defined by the hamiltonian @xmath4 = \int_{0}^{t } d\tau \bigl\{\frac{1}{2 } \bigl[\partial_\tau \phi(\tau)\bigr]^2 + v[\phi(\tau),\tau]\bigr\};\ ] ] where the disorder potential @xmath5 $ ] is gaussian distributed with a zero mean @xmath6 and the @xmath7-correlations : @xmath8 here the parameter @xmath9 describes the strength of the disorder . note that such system is equivalent to the problem of the kardar - parisi - zang ( kpz ) equation describing the growth in time of an interface in the presence of noise @xcite . in what follows we consider the problem in which the polymer is fixed at the origin , @xmath10 and it is free at @xmath11 . in other words , for a given realization of the random potential @xmath12 the partition function of the considered system is : @xmath13 where @xmath14}\ ] ] is the partition function of the system with the fixed boundary conditions , @xmath10 and @xmath15 and where @xmath16 is the total free energy . besides the usual extensive part @xmath17 ( where @xmath18 is the linear free energy density ) , the total free energy @xmath16 of such system is known to contain the disorder dependent fluctuating contribution @xmath19 . in the limit of large @xmath20 the typical value of the free energy fluctuations scales with @xmath20 as @xmath21 ( see e.g. @xcite ) . in other words , the total free energy of the system can be represented as @xmath22 where @xmath23 is a non - universal parameter , which depends on the temperature and the strength of disorder , and @xmath24 is the random quantity which in the thermodynamic limit @xmath25 is described by a non - trivial universal distribution function @xmath26 . note that according to eqs.([3])-([5 ] ) , the trivial self - averaging contribution @xmath27 to the free energy can be eliminated by a simple redefinition of the partition function : @xmath28 so that @xmath29 where @xmath30 for the similar problem with the zero boundary conditions , @xmath31 , the corresponding distribution function was proved to be described by the gaussian unitary ensemble ( gue ) tracy - widom distribution @xcite . in the course of this proof rather efficient bethe ansatz replica technique has been developed @xcite . in particular , in terms of this technique the corresponding multi - point free energy distribution functions have been derived @xcite . recently , the free energy distribution function for the directed polymer problem with the free boundary conditions , eqs.([1])-([4 ] ) , has been obtained @xcite . it was shown that the function @xmath26 is the gaussian orthogonal ensemble ( goe ) tracy - widom distribution . in this paper i would like to present sufficiently simple alternative way of derivation of the same result which does not require rather complicated technique of the fredholm pfaffian described in @xcite . let us introduce the function @xmath32 which gives the probability that the random free energy is bigger that a given value @xmath24 . it will be shown that in the thermodynamic limit , @xmath33 , this function is equal to the fredholm determinant @xmath34 with the kernel @xmath35 which is the goe tracy - widom distribution @xcite . explicitly , @xmath36\ ] ] where @xmath37 is the solution of the panlev ii differential equation , @xmath38 , with the boundary condition @xmath39 . it should be noted that the present paper is rather technical . the main message of this work is not the final result itself ( which is well known anyway ) but the presentation of the general method and new technical tricks used in the derivation . section ii is devoted to the standard reformulation of the considered problem in terms of one - dimensional @xmath0-particle system of quantum bosons with attractive @xmath7-interactions @xcite . here it is shown that the calculation of the the free energy probability distribution function , eq.([9 ] ) , reduces to the summation over all the spectrum of eigenstates of this @xmath0-particle problem . this summation is performed in section iii , where in the thermodynamic limit , @xmath25 , the result , eqs.([10])-([11 ] ) is derived . the concluding remarks and as well as the key points of the calculations are listed in the final section iv . in terms of the partition function @xmath40 , eq.([7 ] ) , the function @xmath41 , eq.([9 ] ) , can be defined as follows : @xmath42 where @xmath43 denotes the averaging over quenched disorder . indeed , substituting here eq.([7 ] ) , we have @xmath44 \\ \nonumber \\ \nonumber & = & \int_{-\infty}^{+\infty } \ ; df ' \ ; p(f ' ) \ ; \theta\bigl(f - f'\bigr)\end{aligned}\ ] ] which coincides with the definition , eq.([9 ] ) . later on we will see that the integration over @xmath45 in the definition of the partition function , eq.([3 ] ) , requires proper regularization at both limits @xmath46 . for that reason it is convenient to represent it in the form of two contributions : @xmath47 thus , taking into account the definition eq.([6 ] ) , we get @xmath48 where @xmath49 using the relations , eqs.([1 ] ) , ( [ 2 ] ) and ( [ 4 ] ) , after simple gaussian averaging we obtain @xmath50 \ ; \exp\bigl(-\beta h_{n } [ \phi_{1 } , \phi_{2 } , ... , \phi_{n } ] \bigr)\ ] ] where @xmath51 \ ; = \ ; \frac{1}{2 } \int_{0}^{t } d\tau \biggl ( \sum_{a=1}^{n } \bigl[\partial_\tau\phi_{a}(\tau)\bigr]^2 - \beta u \sum_{a\not= b}^{n } \delta\bigl[\phi_{a}(\tau)-\phi_{b}(\tau)\bigr ] \biggr)\ ] ] the propagator @xmath52 , eq.([17 ] ) , describes @xmath0 trajectories @xmath53 all starting at zero ( @xmath54 ) , and coming to @xmath0 different points @xmath55 at @xmath11 . one can easily show that @xmath52 can be obtained as the solution of the linear differential equation @xmath56 with the initial condition @xmath57 and the interaction parameter @xmath58 . one can easily see that eq.([19 ] ) is the imaginary - time schrdinger equation @xmath59 with the hamiltonian @xmath60 which describes @xmath0 bose - particles interacting via the _ attractive _ two - body potential @xmath61 . a generic eigenstate of such system is characterized by @xmath0 momenta @xmath62 which are splitted into @xmath63 ( @xmath64 ) `` clusters '' described by continuous real momenta @xmath65 @xmath66 and having @xmath67 discrete imaginary `` components '' ( for details see @xcite ) : @xmath68 with the constraint @xmath69 a generic solution @xmath70 of the schrdinger equation ( [ 19 ] ) with the initial conditions , eq.([20 ] ) , can be represented in the form of the linear combination of the eigenfunctions @xmath71 : @xmath72 \ ; \psi^{(m)}_{{\bf q}}({\bf x } ) { \psi^{(m)}_{{\bf q}}}^{}({\bf 0 } ) \ ; \exp\bigl\{-e_{m}({\bf q } ) t \bigr\}\ ] ] where we have introduced the notation @xmath73 { \boldsymbol \delta}\bigl(\sum_{\alpha=1}^{m } n_{\alpha } \ ; , \ ; n\bigr)\ ] ] and @xmath74 is the kronecker symbol ; note that the presence of this kronecker symbol in the above equation allows to extend the summations over @xmath67 s to infinity . here ( non - normalized ) eigenfunctions are @xcite @xmath75 \ ; \exp\bigl[i \sum_{a=1}^{n } q_{{\cal p}_{a } } x_{a } \bigr]\ ] ] where the summation goes over @xmath76 permutations @xmath77 of @xmath0 momenta @xmath78 , eq.([23 ] ) , over @xmath0 particles @xmath79 ; the normalization factor @xmath80 and the eigenvalues : @xmath81 note that the eigenfunctions , eq.([28 ] ) , are symmetric with respect to permutations of all its arguments @xmath82 , and @xmath83 in this way the problem of the calculation of the free energy probability distribution function , eq.([15 ] ) , reduces to the summation over all the spectrum of the eigenstates of the @xmath0-particle bosonic problem , which is parametrized by the set of both the continuous , @xmath84 , and the discrete @xmath85 degrees of freedom . substituting eqs.([26])-([31 ] ) into eq.([15 ] ) , ( defining @xmath86 , the factor @xmath18 drops out of the further calculations ) we get : @xmath87 \ ; { \boldsymbol \delta}\bigl(\sum_{\alpha=1}^{m } n_{\alpha } \ ; , \ ; n\bigr ) \ ; \|\tilde{c}_{m}({\bf q } , { \bf n})\|^{2 } \ ; i_{k , l } ( { \bf q } , { \bf n } ) \label{32}\end{aligned}\ ] ] where @xmath88 and @xmath89 \times \prod_{a < b}^{k}\biggl[\frac{q_{{\cal p}_a^{(k ) } } - q_{{\cal p}_b^{(k ) } } - i \kappa } { q_{{\cal p}_a^{(k ) } } - q_{{\cal p}_b^{(k)}}}\biggr ] \times \prod_{c < d}^{l}\biggl[\frac{q_{{\cal p}_c^{(l ) } } - q_{{\cal p}_d^{(l ) } } + i \kappa } { q_{{\cal p}_c^{(l ) } } - q_{{\cal p}_d^{(l)}}}\biggr ] \times \\ \nonumber \\ \nonumber & \times & \int_{-\infty < x_{1 } \leq ... \leq x_{k}\leq 0 } dx_{1 } ... dx_{k } \ ; \exp\bigl[i \sum_{a=1}^{k } ( q_{{\cal p}_{a}^{(k)}}-i\epsilon ) x_{a } \bigr ] \\ \nonumber \\ & \times & \int_{0 \leq y_{l } \leq ... \leq y_{1 } < + \infty } dy_{l } ... dy_{1 } \ ; \exp\bigl[i \sum_{c=1}^{l } ( q_{{\cal p}_{c}^{(l)}}+i\epsilon ) y_{c } \bigr ] \label{33}\end{aligned}\ ] ] here the summation over all permutations @xmath77 of @xmath90 momenta @xmath91 over @xmath92 `` negative '' particles @xmath93 and @xmath94 `` positive '' particles @xmath95 are divided into three parts : the permutations @xmath96 of @xmath92 momenta ( taken at random out of the total list @xmath91 ) over @xmath92 `` negative '' particles , the permutations @xmath97 of the remaining @xmath94 momenta over @xmath94 `` positive '' particles , and finally the permutations @xmath98 ( or the exchange ) of the momenta between the group @xmath99 and the group @xmath100 . note also that the integrations both over @xmath79 s and over @xmath101 s in eq.([33 ] ) require proper regularization at @xmath102 and @xmath103 correspondingly . this is done in the standard way by introducing a supplementary parameter @xmath104 which will be set to zero in final results . the result of the integrations can be represented as follows : @xmath105 \times \\ \nonumber \\ \nonumber & \times & \sum_{{\cal p}^{(k ) } } \ ; \ ; \frac{1}{q^{(-)}_{{\cal p}_{1}^{(k ) } } \bigl(q^{(-)}_{{\cal p}_{1}^{(k ) } } + q^{(-)}_{{\cal p}_{2}^{(k)}}\bigr ) ... \bigl(q^{(-)}_{{\cal p}_{1}^{(k ) } } + ... + q^{(-)}_{{\cal p}_{k}^{(k)}}\bigr ) } \prod_{a < b}^{k}\biggl[\frac{q^{(-)}_{{\cal p}_a^{(k ) } } - q^{(-)}_{{\cal p}_b^{(k ) } } - i \kappa } { q^{(-)}_{{\cal p}_a^{(k ) } } - q^{(-)}_{{\cal p}_b^{(k)}}}\biggr ] \times \\ \nonumber \\ & \times & \sum_{{\cal p}^{(l ) } } \ ; \ ; \frac{(-1)^{l}}{q^{(+)}_{{\cal p}_{1}^{(l ) } } \bigl(q^{(+)}_{{\cal p}_{1}^{(l ) } } + q^{(+)}_{{\cal p}_{2}^{(l)}}\bigr ) ... \bigl(q^{(+)}_{{\cal p}_{1}^{(l ) } } + ... + q^{(+)}_{{\cal p}_{l}^{(l)}}\bigr ) } \prod_{c < d}^{l}\biggl[\frac{q^{(+)}_{{\cal p}_c^{(l ) } } - q^{(+)}_{{\cal p}_d^{(l ) } } + i \kappa } { q^{(+)}_{{\cal p}_c^{(l ) } } - q^{(+)}_{{\cal p}_d^{(l)}}}\biggr ] \label{34}\end{aligned}\ ] ] where @xmath106 using the `` magic '' bethe ansatz combinatorial identity @xcite , @xmath107 \ ; = \ ; \frac{1}{\prod_{a=1}^{n } q_{a } } \ ; \prod_{a < b}^{n}\biggl[\frac{q_{a } + q_{b } + i \kappa } { q_{a } + q_{b}}\biggr]\ ] ] ( where the summation goes over all permutations @xmath108 of @xmath0 momenta @xmath109 ) we get : @xmath105 \times \\ \nonumber \\ & \times & \frac{1}{\prod_{a=1}^{k } q^{(-)}_{{\cal p}_{a}^{(k ) } } } \prod_{a < b}^{k}\biggl[\frac{q^{(-)}_{{\cal p}_a^{(k ) } } + q^{(-)}_{{\cal p}_b^{(k ) } } + i \kappa } { q^{(-)}_{{\cal p}_a^{(k ) } } + q^{(-)}_{{\cal p}_b^{(k)}}}\biggr ] \times \frac{(-1)^{l}}{\prod_{c=1}^{l } q^{(+)}_{{\cal p}_{c}^{(l ) } } } \prod_{c < d}^{l}\biggl[\frac{q^{(+)}_{{\cal p}_c^{(l ) } } + q^{(+)}_{{\cal p}_d^{(l ) } } - i \kappa } { q^{(+)}_{{\cal p}_c^{(l ) } } + q^{(+)}_{{\cal p}_d^{(l)}}}\biggr ] \label{36}\end{aligned}\ ] ] further simplification comes from one important property of the bethe ansatz wave function , eq.([28 ] ) . it has such structure that for ordered particles positions ( e.g. @xmath110 ) in the summation over permutations the momenta @xmath78 belonging to the same cluster also remain ordered . in other words , if we consider the momenta , eq.([23 ] ) , of a cluster @xmath111 , @xmath112 , belonging correspondingly to the particles @xmath113 , the permutation of any two momenta @xmath114 and @xmath115 of this _ ordered _ set gives zero contribution . thus , in order to perform the summation over the permutations @xmath98 in eq.([36 ] ) it is sufficient to split the momenta of each cluster into two parts : @xmath116 , where @xmath117 and where the momenta @xmath118 belong to the particles of the sector @xmath99 , while the momenta @xmath119 belong to the particles of the sector @xmath100 . let us introduce the numbering of the momenta of the sector @xmath100 in the reversed order : @xmath120 where @xmath121 and ( s.f . eq.([23 ] ) ) @xmath122 by definition , the integer parameters @xmath123 and @xmath124 fulfill the global constrains @xmath125 in this way the summation over permutations @xmath98 in eq.([36 ] ) is changed by the summations over the integer parameters @xmath123 and @xmath124 : @xmath126 \ ; { \boldsymbol \delta}\bigl(\sum_{\alpha=1}^{m } m_{\alpha}\ ; , \ ; k\bigr ) \ ; { \boldsymbol \delta}\bigl(\sum_{\alpha=1}^{m } s_{\alpha}\ ; , \ ; l\bigr ) \ ; \bigl ( ... \bigr)\ ] ] which allows to lift the summations over @xmath92 , @xmath94 , and @xmath127 in eq.([32 ] ) . in terms of the parameters @xmath123 and @xmath124 the product factors in eq.([36 ] ) are expressed as follows : @xmath128 @xmath129 & = & \prod_{\alpha=1}^{m } \prod_{1\leq r < r'}^{m_{\alpha } } \biggl [ \frac { { q^{\alpha}_{r}}^{(-)}+{q^{\alpha}_{r'}}^{(-)}+i\kappa } { { q^{\alpha}_{r}}^{(-)}+{q^{\alpha}_{r'}}^{(- ) } } \biggr ] \times \prod_{1\leq\alpha<\beta}^{m } \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{m_{\beta } } \biggl [ \frac { { q^{\alpha}_{r}}^{(-)}+{q^{\beta}_{r'}}^{(-)}+i\kappa } { { q^{\alpha}_{r}}^{(-)}+{q^{\beta}_{r'}}^{(- ) } } \biggr ] \\ \nonumber \\ \nonumber \\ \label{45 } \prod_{c < d}^{l } \biggl [ \frac { q^{(+)}_{{\cal p}_c^{(l ) } } + q^{(+)}_{{\cal p}_d^{(l)}}-i \kappa } { q^{(+)}_{{\cal p}_c^{(l ) } } + q^{(+)}_{{\cal p}_d^{(l)}}}\biggr ] & = & \prod_{\alpha=1}^{m } \prod_{1\leq r < r'}^{s_{\alpha } } \biggl [ \frac { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+)}-i\kappa } { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+ ) } } \biggr ] \times \prod_{1\leq\alpha<\beta}^{m } \prod_{r=1}^{s_{\alpha}}\prod_{r'=1}^{s_{\beta } } \biggl [ \frac { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\beta}_{r'}}^{(+)}-i\kappa } { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\beta}_{r'}}^{(+ ) } } \biggr ] \\ \nonumber \\ \nonumber \prod_{a=1}^{k } \prod_{c=1}^{l } \biggl [ \frac { q_{{\cal p}_a^{(k ) } } - q_{{\cal p}_c^{(l ) } } - i\kappa } { q_{{\cal p}_a^{(k ) } } - q_{{\cal p}_c^{(l ) } } } \biggr ] & = & \prod_{1\leq\alpha<\beta}^{m } \biggl\ { \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{s_{\beta } } \biggl [ \frac { q^{\alpha}_{r } - { q^{}}^{\beta}_{r ' } - i\kappa } { q^{\alpha}_{r } + { q^{}}^{\beta}_{r ' } } \biggr ] \times \prod_{r=1}^{s_{\alpha}}\prod_{r'=1}^{m_{\beta } } \biggl [ \frac { { q^{}}^{\alpha}_{r } - q^{\beta}_{r ' } - i\kappa } { { q^{}}^{\alpha}_{r } - q^{\beta}_{r ' } } \biggr ] \biggr\ } \times \\ \nonumber \\ \nonumber \\ \label{46 } & \times & \prod_{\alpha=1}^{m } \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{s_{\alpha } } \biggl [ \frac { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } - i\kappa } { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } } \biggr]\end{aligned}\ ] ] substituting eqs.([41])-([46 ] ) into eq.([36 ] ) , and then substituting the resulting expression into eq.([32 ] ) we obtain @xmath130 \ ; \ ; \prod_{1\leq\alpha<\beta}^{m } \ ; { \cal g}_{\alpha\beta } \bigl({\bf q } , { \bf m } , { \bf s}\bigr ) \biggr\ } \label{47}\end{aligned}\ ] ] where @xmath131 @xmath132 \prod_{r < r'}^{s_{\alpha } } \biggl [ \frac { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+)}-i\kappa } { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+ ) } } \biggr ] \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{s_{\alpha } } \biggl [ \frac { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } - i\kappa } { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } } \biggr]\ ] ] and @xmath133 \prod_{r=1}^{s_{\alpha}}\prod_{r'=1}^{s_{\beta } } \biggl [ \frac { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\beta}_{r'}}^{(+)}-i\kappa } { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\beta}_{r'}}^{(+ ) } } \biggr ] \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{s_{\beta } } \biggl [ \frac { q^{\alpha}_{r } - { q^{}}^{\beta}_{r ' } - i\kappa } { q^{\alpha}_{r } + { q^{}}^{\beta}_{r ' } } \biggr ] \times \prod_{r=1}^{s_{\alpha}}\prod_{r'=1}^{m_{\beta } } \biggl [ \frac { { q^{}}^{\alpha}_{r } - q^{\beta}_{r ' } - i\kappa } { { q^{}}^{\alpha}_{r } - q^{\beta}_{r ' } } \biggr]\ ] ] the product factors in eq.([48 ] ) can be easily expressed it terms of the gamma functions : @xmath134 \ ; = \ ; ( i\kappa)^{m_{\alpha } } \frac { \gamma\bigl ( \frac{1}{2}-\frac{s_{\alpha}-m_{\alpha}}{2 } - \frac{i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) } { \gamma\bigl ( \frac{1}{2}-\frac{s_{\alpha}+m_{\alpha}}{2 } - \frac{i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) } \\ \nonumber \\ \nonumber \\ \prod_{r=1}^{s_{\alpha}}{{q^{}}^{\alpha}_{r}}^{(+ ) } & = & \prod_{r=1}^{s_{\alpha } } \bigl [ { q_{\alpha}}^{(+ ) } + \frac{i\kappa}{2}(m_{\alpha}+s_{\alpha}+1 ) - i\kappa r \bigr ] \ ; = \ ; ( -i\kappa)^{s_{\alpha } } \frac { \gamma\bigl ( \frac{1}{2}-\frac{m_{\alpha}-s_{\alpha}}{2 } + \frac{i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) } { \gamma\bigl ( \frac{1}{2}-\frac{m_{\alpha}+s_{\alpha}}{2 } + \frac{i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) } \label{51}\end{aligned}\ ] ] @xmath135 & = & 2^{-(m_{\alpha}-1 ) } \frac { \gamma\bigl ( m_{\alpha}-s_{\alpha } - \frac{2i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) \gamma\bigl ( 1-\frac{m_{\alpha}+s_{\alpha}}{2 } - \frac{i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) } { \gamma\bigl ( \frac{m_{\alpha}-s_{\alpha}}{2 } - \frac{i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) \gamma\bigl ( 1 - s_{\alpha } - \frac{2i{q_{\alpha}}^{(-)}}{\kappa } \bigr ) } \\ \nonumber \\ \nonumber \\ \label{53 } \prod_{r < r'}^{s_{\alpha } } \biggl [ \frac { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+)}-i\kappa } { { { q^{}}^{\alpha}_{r}}^{(+)}+{{q^{}}^{\alpha}_{r'}}^{(+ ) } } \biggr ] & = & 2^{-(s_{\alpha}-1 ) } \frac { \gamma\bigl ( s_{\alpha}-m_{\alpha } + \frac{2i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) \gamma\bigl ( 1-\frac{m_{\alpha}+s_{\alpha}}{2 } + \frac{i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) } { \gamma\bigl ( \frac{s_{\alpha}-m_{\alpha}}{2 } + \frac{i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) \gamma\bigl ( 1 - m_{\alpha } + \frac{2i{q_{\alpha}}^{(+)}}{\kappa } \bigr ) } \\ \nonumber \\ \nonumber \\ \label{54 } \prod_{r=1}^{m_{\alpha}}\prod_{r'=1}^{s_{\alpha } } \biggl [ \frac { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } - i\kappa } { q^{\alpha}_{r } - { q^{}}^{\alpha}_{r ' } } \biggr ] & = & \frac { \gamma\bigl(1 + m_{\alpha } + s_{\alpha}\bigr ) } { \gamma\bigl(1 + m_{\alpha}\bigr ) \gamma\bigl(1 + s_{\alpha}\bigr)}\end{aligned}\ ] ] substituting the above expressions into eq.([48 ] ) and using the standard relations for the gamma functions , @xmath136 for the factor @xmath137 , eq.([48 ] ) , we get @xmath138 similar calculations for the factor @xmath139 yield the following expression @xmath140 \ , \gamma \bigl [ 1 - \frac{m_{\alpha } + m_{\beta } + s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl({q_{\alpha}}^{(- ) } + { q_{\beta}}^{(-)}\bigr ) \bigr ] } { \gamma \bigl [ 1 - \frac{m_{\alpha } - m_{\beta } + s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl({q_{\alpha}}^{(- ) } + { q_{\beta}}^{(-)}\bigr ) \bigr ] \ , \gamma \bigl [ 1 + \frac{m_{\alpha } - m_{\beta } - s_{\alpha } - s_{\beta}}{2 } -\frac{i}{\kappa}\bigl({q_{\alpha}}^{(- ) } + { q_{\beta}}^{(-)}\bigr ) \bigr ] } \times \\ \nonumber \\ \nonumber \\ \nonumber & \times & \frac { \gamma \bigl [ 1 - \frac{m_{\alpha } + m_{\beta } - s_{\alpha } - s_{\beta}}{2 } + \frac{i}{\kappa}\bigl({q_{\alpha}}^{(+ ) } + { q_{\beta}}^{(+)}\bigr ) \bigr ] \ , \gamma \bigl [ 1 - \frac{m_{\alpha } + m_{\beta } + s_{\alpha } + s_{\beta}}{2 } + \frac{i}{\kappa}\bigl({q_{\alpha}}^{(+ ) } + { q_{\beta}}^{(+)}\bigr ) \bigr ] } { \gamma \bigl [ 1 - \frac{m_{\alpha } + m_{\beta } + s_{\alpha } - s_{\beta}}{2 } + \frac{i}{\kappa}\bigl({q_{\alpha}}^{(+ ) } + { q_{\beta}}^{(+)}\bigr ) \bigr ] \ , \gamma \bigl [ 1 - \frac{m_{\alpha } + m_{\beta } - s_{\alpha } + s_{\beta}}{2 } + \frac{i}{\kappa}\bigl({q_{\alpha}}^{(+ ) } + { q_{\beta}}^{(+)}\bigr ) \bigr ] } \times \\ \nonumber \\ \nonumber \\ \nonumber & \times & \frac { \gamma \bigl [ 1 + \frac{m_{\alpha } + m_{\beta } + s_{\alpha } + s_{\beta}}{2 } + \frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] \ , \gamma \bigl [ 1 + \frac{-m_{\alpha } + m_{\beta } + s_{\alpha } - s_{\beta}}{2 } + \frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] } { \gamma \bigl [ 1 + \frac{-m_{\alpha } + m_{\beta } + s_{\alpha } + s_{\beta}}{2 } + \frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] \ , \gamma \bigl [ 1 + \frac{m_{\alpha } + m_{\beta } + s_{\alpha } - s_{\beta}}{2 } + \frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] } \times \\ \nonumber \\ \nonumber \\ \label{59 } & \times & \frac { \gamma \bigl [ 1 + \frac{m_{\alpha } + m_{\beta } + s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] \ , \gamma \bigl [ 1 + \frac{m_{\alpha } - m_{\beta } - s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] } { \gamma \bigl [ 1 + \frac{m_{\alpha } + m_{\beta } - s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr ] \ , \gamma \bigl [ 1 + \frac{m_{\alpha } - m_{\beta } + s_{\alpha } + s_{\beta}}{2 } -\frac{i}{\kappa}\bigl(q_{\alpha } - q_{\beta}\bigr ) \bigr]}\end{aligned}\ ] ] redefining @xmath141 with @xmath142 the normalization factor @xmath143 , eq.([47a ] ) , can be represented as follows : @xmath144 \times \det \biggl [ \frac{1 } { \lambda\bigl(m_{\alpha}+s_{\alpha}\bigr ) - ip_{\alpha } + \lambda\bigl(m_{\beta}+s_{\beta}\bigr ) + ip_{\beta } } \biggr]_{\alpha,\beta=1, ... ,m}\end{aligned}\ ] ] where we have used the cauchy double alternant identity @xmath145_{\alpha,\beta=1, ... m}\ ] ] with @xmath146 and @xmath147 . after rescaling , eq.([60 ] ) , for the exponential factor in eq.([47 ] ) we find @xmath148 the cubic exponential term can be linearized using the airy function relation @xmath149 \ ; = \ ; \int_{-\infty}^{+\infty } dy_{\alpha } \ ; \operatorname{ai}(y_{\alpha } ) \ ; \exp\bigl[\lambda ( m_{\alpha}+s_{\alpha } ) \ , y_{\alpha } \bigr]\ ] ] substituting eqs.([65]),([64 ] ) and ( [ 62 ] ) into eq.([47 ] ) , and redefining @xmath150 , we get @xmath151 \times \\ \nonumber \\ & \times & \det \hat{k}\bigl[(\lambda m_{\alpha},\ , \lambda s_{\alpha } , \ , p_{\alpha } ) ; ( \lambda m_{\beta } , \ , \lambda s_{\beta } , \ , p_{\beta})\bigr]_{\alpha,\beta=1, ... ,m } \times \prod_{1\leq\alpha<\beta}^{m } \ ; { \cal g}_{\alpha\beta}\bigl(\frac{{\bf p}}{\lambda } , \ ; { \bf m } , \ ; { \bf s}\bigr ) \biggr\ } \label{66}\end{aligned}\ ] ] where @xmath152 \ ; = \ ; \frac{1 } { \lambda m + \lambda s - ip + \lambda m ' + \lambda s ' + ip'}\ ] ] the crucial point of the further calculations is the procedure of taking the thermodynamic limit @xmath153 . first of all one can easily note that according to eqs.([60 ] ) and ( [ 66 ] ) , it is the parameters @xmath154 which remain finite in the limit @xmath153 . in other words , all the parameters @xmath65 which are not multiplied by @xmath155 ( e.g. in the expressions for @xmath156 and @xmath139 , eqs([58 ] ) and ( [ 59 ] ) ) have to be taken to zero in this limit . simultaneously , the summations over @xmath123 and @xmath124 have to be performed . the general algorithm of such summation is in the following . let us consider the example of the sum of a general type : @xmath157\ ; \phi\bigl [ { \bf p},\ ; \frac{{\bf p}}{\lambda } , \ ; \lambda { \bf n } ; \ ; { \bf n } \bigr]\ ] ] where @xmath158 is a function which depend both of @xmath159 s and @xmath67 s ( which are not multiplied by @xmath155 ) . the summations in the above example can be represented in terms of the integrals in the complex plane : @xmath160\ ; \phi\bigl [ { \bf p},\ ; \frac{{\bf p}}{\lambda } , \ ; \lambda { \bf z } ; \ ; { \bf z } \bigr]\ ] ] where the integration goes over the contour @xmath161 shown in fig.1(a ) . shifting the contour to the position @xmath162 shown in fig.1(b ) ( assuming that there is no contribution from @xmath163 ) , and redefining @xmath164 , in the limit @xmath153 we get : @xmath165\ ; \lim_{\lambda\to\infty } \phi\bigl [ { \bf p},\ ; \frac{{\bf p}}{\lambda } , \ ; { \bf z } ; \ ; \frac{{\bf z}}{\lambda } \bigr]\ ] ] where the parameters @xmath166 , @xmath167 and @xmath168 remain finite in the limit @xmath153 . ; ( b ) the deformed contour @xmath162 ; , width=453 ] [ figure1 ] let us consider now the summations in eq.([66 ] ) . here the double sum can be represented as follows : @xmath169 thus , according to the above summation algorithm , we get @xmath170 \lim_{\lambda\to\infty } f\bigl(\frac{{\bf z}_{1}}{\lambda } ; \frac{{\bf z}_{2}}{\lambda}\bigr)\ ] ] where the rescaled integration parameters @xmath171 and @xmath172 remain finite in the limit @xmath153 . finally , taking into account the gamma function properties , @xmath173 and @xmath174 , for the factors @xmath137 and @xmath139 , eqs.([58])-([59 ] ) , we easily find @xmath175 and @xmath176 where @xmath177 thus , in the limit @xmath153 the expression for the probability distribution function , eq.([66 ] ) , takes the form of the fredholm determinant @xmath178 \bigl ( 1 + \frac{{z_{1}}_{\alpha}}{{z_{2}}_{\alpha } + i p_{\alpha}^{(- ) } } \bigr ) \bigl ( 1 + \frac{{z_{2}}_{\alpha}}{{z_{1}}_{\alpha } - i p_{\alpha}^{(+ ) } } \bigr ) \ ; \mbox{\large e}^{({z_{1}}_{\alpha } + { z_{2}}_{\alpha } ) y_{\alpha } } \biggr ] \\ \nonumber \\ \nonumber & \times & \det \hat{k}\bigl[({z_{1}}_{\alpha},{z_{2}}_{\alpha},p_{\alpha } ) ; ( { z_{1}}_{\beta},{z_{2}}_{\beta},p_{\beta})\bigr]_{\alpha,\beta=1, ... ,m } \\ \nonumber \\ & = & \det\bigl[1 \ , - \ , \hat{k } \bigr ] \label{76}\end{aligned}\ ] ] with the kernel @xmath179 \ ; = \ ; \frac{1 } { { z_{1 } } + { z_{2 } } - ip + { z_{1 } } ' + { z_{2 } } ' + ip'}\ ] ] in the exponential representation of this determinant we get @xmath180\ ] ] where @xmath181 \bigl ( 1 + \frac{{z_{1}}_{\alpha}}{{z_{2}}_{\alpha } + i p_{\alpha}^{(- ) } } \bigr ) \bigl ( 1 + \frac{{z_{2}}_{\alpha}}{{z_{1}}_{\alpha } - i p_{\alpha}^{(+ ) } } \bigr ) \ ; \mbox{\large e}^ { ( { z_{1}}_{\alpha } + { z_{2}}_{\alpha } ) y_{\alpha } } \biggr ] \times \\ \nonumber \\ & \times & \prod_{\alpha=1}^{m } \biggl [ \frac{1 } { { z_{1}}_{\alpha } + { z_{2}}_{\alpha } - i p_{\alpha } + { z_{1}}_{\alpha + 1 } + { z_{2}}_{\alpha + 1 } + i p_{\alpha + 1 } } \biggr ] \label{79}\end{aligned}\ ] ] here , by definition , it is assumed that @xmath182 ( @xmath183 ) and @xmath184 . substituting @xmath185\ ] ] into eq.([79 ] ) , we obtain @xmath186\ ] ] where , by definition , @xmath187 , and @xmath188 \bigl ( 1 + \frac{z_{1}}{z_{2 } + i p^{(- ) } } \bigr ) \bigl ( 1 + \frac{z_{2}}{z_{1 } - i p^{(+ ) } } \bigr ) \ ; \mbox{\large e}^{(z_{1 } + z_{2 } ) y_{\alpha}}\ ] ] simple calculations yield : @xmath189 \ ; \delta(y)\end{aligned}\ ] ] taking the limit @xmath190 we find : @xmath191 substituting this result into eq.([81 ] ) we obtain @xmath192\ ] ] in other words , the free energy distribution function of our problem is given by the fredholm determinant , @xmath193\ ] ] with the kernel @xmath194 which is the goe tracy - widom distribution @xcite . in this paper we have presented sufficiently simple derivation of the goe tracy - widom distribution function for the free energy fluctuations in random directed polymers with free boundary conditions . the main message of this somewhat technical work is not the final result itself ( which is not new anyway ) , but the demonstration of the efficiency of the general method and new technical tricks used in the derivation . by mapping the original problem to the @xmath0-particle quantum boson system with attractive interactions the derivation is done in the framework of the integer replica series summations and the bethe ansatz formalism for the quantum boson system . the key technical tricks of presented calculations includes the following points . first of all , to make the integration over particle coordinates of the bethe ansatz propagator well defined one has to introduce proper regularization at @xmath46 which requires formal splitting the partition function into two parts : the one in the positive particles coordinates sector ( up to @xmath103 ) and another one in the negative particles coordinates sector ( down to @xmath102 ) , eqs.([14])-([15 ] ) . next is the `` magic '' bethe ansatz combinatorial identity , eq.([35 ] ) , which allows to perform the summation over the momenta permutations and `` disentangle '' sophisticated products containing in the bethe ansatz propagator . one more trick is the reformulation of the summation over permutations of the momenta between the positive and the negative particles positions sectors in terms of the series summations , eq.([41 ] ) , which allows to represent the probability distribution function in terms of the problem of the series summations , eq.([47 ] ) . finally , the crucial point of the considered derivation is the procedure of the series summations in the thermodynamic limit @xmath33 . in this limit , due to the integral representation of the series , eqs.([68])-([70 ] ) , one obtains dramatic simplifications of some factors , eqs.([73])-([74 ] ) , in the expression for the probability distribution function , which allows to represent it in the form of the fredholm determinant , eq.([76 ] ) . hopefully , the experience obtained in the presented calculations would help to solve more serious long standing problems of this scope , such as the distribution function of the directed polymer s end point fluctuations or the statistical properties of the free energy fluctuations at different times . an essential part of this work was done during my visit to the fields institute , toronto , in the spring of 2011 in the framework of the program `` dynamics and transport in disordered systems '' . i would like to thank jeremy quastel and kostya khanin for numerous illuminating discussions which were crucial for the progress in these somewhat complicated calculations .	the distribution function of the free energy fluctuations in one - dimensional directed polymers with free boundary conditions is derived by mapping the replicated problem to the @xmath0-particle quantum boson system with attractive interactions . it is shown that in the thermodynamic limit this function is described by the universal tracy - widom distribution of the gaussian orthogonal ensemble .
sn 1997ef in ugc4107 was recognised as a peculiar and interesting object as soon as its first spectra were taken by the harvard - cfa team ( garnavich et al . 1997a , b , c ) . the spectra displayed very broad features , quite unlike those of any other sn known to date , so that it was not even clear whether what was observed was an absorption or an emission spectrum . continued observation revealed a light curve typical of a sn deriving from a compact object ( a sn ia or a sn ic ) , but much broader than the templates for both of these classes . later spectra showed more resolved spectral lines , clarifying that the broad features were actually very extended absorption blends , and that the sn was therefore in the photospheric epoch . lines of ca ii , o i and fe ii were strong , and the si ii 6347 , 6371 line appeared to be comparatively weak , so the sn was classified as type ic , as also supported by the overall similarity with the spectra of other sne ic such as sn 1994i ( filippenko 1997 ; millard et al . 1999 ) . typical models for a low mass , low kinetic energy sn ic ( nomoto et al . 1994 ; iwamoto et al . 1994 ) gave good fits to the light curve , but the synthetic spectra computed using such models , while yielding the correct types of spectral lines , completely failed to reproduce the observed large line width , which we estimated as at least 20000 km s@xmath0for some of the strongest lines ( iwamoto et al . 2000 , hereafter paper i ) . soon after sn 1997ef , however , the extraordinarily bright and powerful type ic sn 1998bw was discovered as the optical counterpart of grb980425 ( galama et al . a great deal of study , both observational and theoretical , was and still is being devoted to this sn , which was soon realised to be a massive type ic event of exceptionally large kinetic energy ( @xmath6 ergs , iwamoto et al . 1998 ; woosley , eastman & schmidt 1999 ; branch 2000 ) , where the link with the grb may lie in a rather asymmetric explosion ( macfadyen & woosley 1999 , hflich et al . 1999 , khokhlov et al . 1999 , maeda et al . it was only natural then also to analyse sn 1997ef as a powerful ` hypernova ' . paper i presents a detailed study of the light curve and the spectrum using two models , one for a ` normal ' sn ic ( model co60 : m@xmath7= 4.4m@xmath2 , ke = 10@xmath8 erg ) and another for a ` hypernova ' ( model co100 : m@xmath7= 7.6m@xmath2 , @xmath9 erg ) . both models have a @xmath5ni mass of 0.15m@xmath2 . in paper i we showed that both models reproduce the light curve reasonably well , but only the more massive and energetic model co100 can also produce reasonable synthetic spectra , and therefore they opted for the ` hypernova ' model and suggested that sn 1997ef may be a somewhat less extreme case than sn 1998bw . a possible connection with a grb has been searched for in archives for sn 1997ef . the result was that grb971115 could be compatible with sn 1997ef in position and time of occurrence ( wang & wheeler 1998 ) , but the statistical significance of the correlation is much weaker than for sn 1998bw and grb980425 . in this paper , we concentrate on the spectroscopy of sn 1997ef . this is interesting in many respcts . first , it is important that we not only fit the spectra and their evolution using one explosion model , but that we also understand which lines are present in the spectrum . since we have available a sufficiently large number of observed spectra , we can follow the spectral evolution and thus probe different depths in the ejecta and verify the composition as a function of velocity , and also describe how the spectrum changes as a function of luminosity as the sn moves along the light curve . we use for this purpose the monte carlo ( mc ) code originally described in mazzali & lucy ( 1993 ) and further improved and discussed by lucy ( 1999 ) and mazzali ( 2000 ) . the code includes the effect of photon branching , which must be very important in a situation where the high expansion velocities lead to extensive line blanketing . also , since the code uses the bolometric luminosity as input , we can compare the required @xmath10 values with those of a light curve calculation relaxing the assumption @xmath11 made in paper i and thus improve the comparison between model @xmath12 from light curve calculations and observed @xmath13 mags . furthermore , we use spectrum synthesis to date the earliest spectra , and thus to determine with some accuracy the time of explosion , which is important for both the grb connection and the light curve calculation . finally , having noticed that even synthetic spectra obtained from co100 do not give wonderfully good fits to the observed spectra , we adopt the ` inverse ' approach and try to determine the density and abundance distribution with velocity by looking for a best fit to the spectra . this method was pioneered by branch ( 2000 ) , who suggested for sn 1997ef a much flatter density dependence ( @xmath14 ) than what is predicted by model co100 ( where the density index is @xmath15 . consequently , for sn 1997ef branch ( 2000 ) obtains a much larger ejecta mass ( 6m@xmath2 above @xmath16 km s@xmath0 ) and kinetic energy ( @xmath17erg above the same velocity ) . since we study several spectra we can control @xmath18 as a function of @xmath19 , and hence build our own @xmath20 . we do confirm that the spectra require a flatter density dependence , but we show that this is only necessary at high velocity ( @xmath21 km s@xmath0 ) . examining spectra at advanced epochs , we also show that the predicted density ` hole ' at low velocities is not compatible with observations . we therefore derive new values for both m@xmath7and @xmath22 . we take this process one step further by computing a synthetic light curve from our derived @xmath20 , and show that this gives a very good fit to the observed light curve of sn 1997ef . in the rest of this paper we first describe how we proceeded to date the spectra , then discuss the analysis of 6 spectra , reviewing the results and their implications . then we review the properties of the model for sn 1997ef as we derived it , and discuss how it differs from model co100 , which was computed to fit the light curve , not the spectra . we also show the synthetic light curve obtained from our ad hoc model for @xmath20 . finally , we review and discuss our findings , including possible reasons why sn 1997ef was a hypernova but no grb was apparently linked to it . we have selected for modelling 6 spectra of sn 1997ef obtained by the harvard - cfa group ( garnavich et al . in prep . ; see paper i , figs . 8 and 9 ) . we have chosen well exposed , high s / n spectra , with a large wavelength coverage , and tried to sample the light curve as uniformly as possible . the first three spectra are from epochs around maximum ( which was not observed either spectroscopically or photometrically but was probably reached around dec 10 , 1997 at @xmath23 . the dates of the spectra are nov 29 , which is very soon after discovery , dec 5 and dec 17 . all the spectra show very broad lines . they are all early enough that they are sensitive to changes in the kinetic energy and to the assumed epoch of outburst . the next two spectra have dates dec 24 and jan 1 , which is around the time when the sn enters the tail of the light curve . the lines in these spectra becoming progressively narrower , indicating that the high velocity part of the ejecta is becoming optically thin . the final spectrum has date jan 26 , 1998 , which is well on the light curve tail . line velocities are quite small , suggesting that the spectrum is formed deep in the ejecta . one of the challenges for a good light curve calculation is to follow the evolution of the photospheric velocity with time . even model co100 was not perfect in this respect ( cf . paper i , figure 7 ) . since the epoch of maximum was not observed , light curve models are not very tightly constrained . since the rise time to maximum depends on the structure of the model and the abundance distribution , the range of allowed parameters could be effectively costrained if at least the time of explosion could be determined , albeit with some uncertainty . this could be attempted using spectrum synthesis and the measured photospheric velocity as determined in paper i from the velocity of the si ii line . it could be expected that this line is formed somewhat above the photosphere at very early epochs , like in sne ia ( e.g. sn 1994d , patat et al . 1996 ) . since our synthetic spectra are computed using the luminosity @xmath10 , the epoch @xmath24 and the photospheric velocity @xmath25 as input , an appropriate value of @xmath24 must be found that gives a photospheric radius @xmath26 which , when combined with @xmath10 , produces temperatures in agreement with the observed spectrum . in particular , although the continuum is not well determined in the very complex spectra of sn 1997ef , spectral lines serve a double role : they depend on the temperature ( and hence on @xmath27 and @xmath10 ) through their ratios , and on @xmath25 through their displacement . not many lines are visible in the earliest spectra , but at least si ii 6346 , 6371 and the broad complex which can most likely be identified as o i 7774 , o i8447 and the ca ii ir triplet could be used to guide the calculation of the synthetic spectra . models for nov 29 show that acceptable fits are found for @xmath28 days and @xmath29 km s@xmath0 . greater epochs require smaller velocities , so that @xmath27 always has a value of about @xmath30 cm . this radius , combined with the value of @xmath10 , which depends almost entirely on the assumed distance , gives a temperature appropriate to get a good overall fit to the spectrum . if we assume e.g. @xmath31 days , then we find a good overall fit to the spectrum for @xmath32 km s@xmath0 , which is too small and is reflected on the line shift . the opposite happens if @xmath33 days . in order to strenghten our result , we applied the same technique to the dec 5 spectrum , and we found acceptable fits for @xmath34 days and @xmath35 km s@xmath0 . therefore , the two epochs give consistent results , and we can estimate the time of outburst to have been nov 2022 . since the fits obtained for @xmath36 days on nov 29 and @xmath37 days on dec 5 were particularly good , we selected nov 20 as our reference epoch , also for the comparison with the light curve and the photospheric velocities . we will show the fits in the next sections , when discussing the various epochs . in figure 1 we show as an example the nov 29 spectrum and two synthetic spectra computed as above for @xmath36 days ( dashed line ) and @xmath31 days ( dotted line ) , respectively , using model co100 . the difference in velocity is clearly visible . these results depend on the selection made for two other quantities , distance and overall ejecta mass , as discussed by mazzali & schmidt ( 2000 ) in the context of type ia sne . for all models , we adopted a distance modulus of 33.63 , i.e. a distance of 53.1 mpc , as estimated from a recession velocity of 3450 km s@xmath0(p . garnavich , 2000 , private communication ) and @xmath38 km s@xmath0mpc@xmath0 . we assume no extinction ( @xmath39 , which is justified by the absence of a narrow interstellar na i d line from the spectrum of sn 1997ef at all phases . as for the ejected mass , we already showed in paper i that this must be large in order to fit both the light curve and the spectra . therefore in this paper we use the solution found in paper i , model co100 , as a reference . this is to our knowledge the earliest spectrum available of sn 1997ef . although it is somewhat noisy and it does not extend very far to the red , it nevertheless includes most of the main features , which made sn 1997ef such a peculiar and interesting object . the spectrum ( figure 2 ) has essentially no continuum , and it is characterised by four broad minima , at roughly 3900 , 4700 , 6000 and 7500 . the minima are separated by three rather sharp peaks , at 4300 , 5200 and 6300 , respectively . the minima are so broad and the peaks so strong that the spectrum could be confused for an emission spectrum . both branch ( 2000 ) and paper i showed that the real nature of the spectrum is a superposition of absorption lines . we already pointed out that the ` emission peaks ' are the result of photon travel in a medium of large line opacity . photons redshift their way through the envelope until they find a portion of the spectrum where line opacity is low because of the intrinsic distribution in wavelength of the atomic transitions . we call such a region a ` line - free window ' . a large number of photons escape through these regions - all the photons coming from the optically thick region immediately to the blue , and this leads to a large flux and to the observed ` peaks ' . while the width of the peak depends solely on the width of the line - free window , its strength depends both on the flux at the lower boundary ( the photosphere ) and on the width of the blanketing region which feeds photons to it : the broader the blanketing region , the more photons must reach the window to escape , resulting in a higher peak flux . this can be easily seen by inspection of the spectrum . in paper i we showed that a synthetic spectrum based on model co100 gives a reasonably good reproduction of the observations , and allows us to identify the four absorptions as follows : 3900 : ca ii h&k + ni ii , co ii and si iilines ; 4700 : mostly fe ii lines ; 6000 : si ii 6347 , 6371 plus weaker si ii lines and na i d further to the blue ; 7500 : o i 7774 . in spectra computed with model co100 the o i 7774 and the ca ii ir triplet are well separated , but unfortunately the observed spectrum does not extend far enough to the red to show the ca ii region . with reference to that synthetic spectrum , which is shown again here as the dashed line in figure 2 , we notice that none of the absorptions extend quite far enough to the blue . for the two bluer troughs , which are blends of very many lines , this may possibly be due to the lack of lines resulting from neglecting some element or not using a sufficiently complete line list , but this seems unlikely for the two redder troughs , where a few very strong lines dominate . the synthetic spectrum has @xmath36 days , @xmath40 ( erg s@xmath0 ) , @xmath41 km s@xmath0 . the observed @xmath13 magnitude ( 16.7 ) and the synthetic one ( 16.75 ) agree quite well , but the spectrum has a large bolometric correction ( 0.28 ) , which indicates that the assumption @xmath42 made in paper i when fitting the light curve is at least weak . the observed si ii line velocity is 19000 km s@xmath0 , and the core of the si ii absorption is well reproduced even using model co100 . therefore , in order to improve the synthetic spectrum , we modified the density only above 20000 km s@xmath0 . while the original co100 model has @xmath43 in the outer part , we made the outer density radial dependence flatter . as a consequence , high velocity absorption becomes stronger . of course , this happens first in the strongest lines , the ca ii h&k doublet and ir triplet first and foremost . the strength of these lines increases dramatically , because they come from low excitation levels which are still highly populated even at the very low densities and temperatures of the outer envelope ( @xmath44k ) , and so these lines increase greatly in width . next come the si ii and o i lines . for these lines , the drop in the level population with radius is much steeper , since they come from levels on average about 10ev above the ground state , and so they do become broader , but not so noticeably , and their cores do not shift much towards the blue . all other lines , strong and weak are similarly affected , to a higher or lesser degree depending on the ionisation and excitation of the levels . since we wanted to introduce as little change from model co100 as possible , we choose the steepest density profile which gave a good fit to the spectrum . this was obtained with a density power law index @xmath45 at @xmath21 km s@xmath0 . when this change is introduced the ca ii h&k doublet , the si ii line and the o i7774 line broaden to the point that the observed absorption troughs are very well reproduced . further to the red , in the unobserved part of the spectrum , the ca ii ir triplet is predicted to broaden and blend with o i 7774 . the contribution of o i 8446 , which is also a strong line , is nevertheless small compared to that of ca ii . the blending of o i 7774 and the ca ii ir triplet is an important discriminant for the density structure at high velocities . fortunately , the dec . 5 spectrum covers that region . the blue sides of the peaks at 4300 and 5000 are also fitted much better , as the increased density causes the emission to set in at higher velocities . the only region where no improvement is seen is near 4500 . the fe ii and co ii lines causing the absorption near 4750 are not sufficiently strong to extend to the blue , and there are not enough strong lines that absorb near 44004500 . it is possible that we are either missing lines or ignoring some relevant element . interestingly , branch ( 2000 ) , who used a code that is very efficient for identifying lines , seemed to have the same problem . the modification to the outer density profile has only a small influence on the total mass , since it adds only about 0.25m@xmath2 in the outer ejecta , but it does lead to a significant increase in the overall kinetic energy , since all the mass is added to the high velocity shells , where it represents a dramatic increase , of 12 orders of magnitude . the kinetic energy above 20000 km s@xmath0 increases from @xmath46 to about @xmath47erg , and the total kinetic energy increases from @xmath48 to @xmath3erg . this is quite a large change , but it is still smaller than the value adopted for sn 1998bw . also , both the outer slope and the revised @xmath22 are , respectively , steeper and smaller than the values suggested by branch ( 2000 ) . we will discuss the implications of these revised values later . an increase in @xmath22 only makes sn 1997ef more like a hypernova . the best fitting ad hoc model has the same input parameters as the spectrum computed using co100 . the increased outer density makes the temperature only slightly higher . the synthetic spectrum has @xmath49 , @xmath50 and @xmath51 , so that @xmath52 . therefore comparing @xmath53 with @xmath13 is misleading , since @xmath13 rises more rapidly than @xmath53 . finally , a word of comment about the presence of he in the spectrum . in sn 1998bw , he i 10830 was almost certainly observed in absorption at early epochs ( patat et al . 2000 ) . its presence could be understood if the he shell was not completely lost before the explosion . unfortunately , ir spectra are not available for sn 1997ef , so the best place to look for he is he i 5868 . this line is very close to na i d , which is rather strong and contributes to the broad 6000 trough with an absorption near 5450 . a small he mass does not produce a strong line unless very large departure coefficients are assumed , and in any case a strong he i line is not necessary to fit the spectrum . therefore , although we can not be final , we do not think he i 5868 is present in the nov 29 spectrum . the second spectrum in our analysis has a good s / n and a very good wavelength coverage , extending from 3100 to 8700 . it is actually quite similar to the 29 nov . one , but more detail is visible ( figure 3 ) , e.g. the structure of the complex 6000 absorption and , very importantly , the ca ii ir triplet is visible , and clearly blended with o i 7774 , which is a strong argument in favour of a ` flat ' outer density distribution , as discussed above . that the feature measured at 7500 is a blend is clear , given the presence of two distinct absorptions , whose wavelengths match those of the two component multiplets , as shown also by the spectrum computed with model co100 . the synthetic spectrum based on model co100 was shown in paper i and again here as the dashed line in figure 3 . it has parameters @xmath37 days , @xmath54 ( erg s@xmath0 ) , @xmath55 km s@xmath0 . both @xmath10 and @xmath27 have values similar to those on nov . 29 , therefore the temperatures are similar and so are the two spectra , although the photosphere is quite a bit deeper on dec . this is shown also by the si ii line velocity , which is reduced to 13000 km s@xmath0 . the observed @xmath13 mag on dec 5 was 16.5 , and the co100 spectrum reproduces it well ( @xmath56 ) . the problem with that model is essentially the same as that discussed for the nov 29 model : all troughs do not extend far enough to the blue . furthermore , o i 7774 and the ca ii ir triplet do not blend anywhere near as much as they should when the original model is used . in that spectrum the ca ii ir triplet extends only to about 7850 , i.e. a ca iivelocity of @xmath57 km if the ca ii ir triplet is to blend with the o i line , it must reach at least 7650 in the blue wing , i.e. @xmath58 km s@xmath0 . this is consistent with the modified density structure adopted for the nov 29 spectrum . since @xmath25 is much lower than the values at which the density has been modified ( @xmath59 km s@xmath0 ) , we can use the same density structure as for nov 29 . the spectrum we computed for the modified structure and the same input parameters as above is shown in figure 3 . the improvements are apparent , and they concern mostly the ca ii ir triplet + o i 7774 feature in the red and the blue region near 4000 , which is impoved by the strengthening and broadening of ca ii h&k . improvements in the trough at 4700 are again only marginal , as are those to the si ii line region . surprisingly perhaps , the si ii line does not become broader , but this can be understood since the high velocity regions where the density has been enhanced are far above the photosphere and are of too low density now to cause much si ii absorption . the situation is made more complicated by the apparent development of an absorption shoulder near 5900 . we can not offer a clear identification for this shoulder , except for several lines of o i at 6157 , but these are not very strong . the shoulder was not present on nov 29 , nor is it on dec 17 . another , stronger shoulder is visible near 5750 . this is where si ii 5958 , 5979falls , and the synthetic spectra show that this doublet , combined with na i d further to the red , is strong enough to produce a shoulder stronger than the observed one without resorting to he i. therefore we can confirm the non - detection of he i 5876 . note that the si abundance in this model was lower than on nov 29 . this was necessary to fit the core of the si ii 6347 , 6371 line . even the modified density does not give a perfect synthetic spectrum , but the excellent behaviour in the red confirms that we are adopting the correct measures when using an extended outer envelope . if a density law flatter than @xmath60 is adopted in an attempt to improve the si ii line two problems emerge : 1 ) o i 7774 extends too far to the blue and 2 ) the increasing strength of ca ii h&k causes too much blocking of the uv spectrum and , consequently , a synthetic peak at 4500 much stronger than what is observed . these problems would be even worse in the nov 29 spectrum , and so we chose to retain the @xmath60 solution as a compromise and suggest that some alternative line may be the reason for at least the shoulder at 5900 . our best - fitting spectrum has @xmath61 , @xmath62 and again a large @xmath63 , so that @xmath64 . the increased @xmath65 is apparent if one looks at the ratio of the two main peaks , at 4300 and 5200 . therefore , @xmath53 is significantly smaller than @xmath13 at this epoch , which is near maximum in the @xmath13 light curve but appears to be still very much on the rising branch of the bolometric curve . this third spectrum has again a limited wavelength coverage , but it is the first one available after dec 5 , and it is therefore useful to study the evolution of the sn immediately following @xmath13 maximum . the spectrum , which is displayed in figure 4 has @xmath66 . the overall aspect of the spectrum is similar to the previous two epochs , which indicates that the temperature conditions have probably not changed greatly . however , closer inspection , and in particular a wavelength comparison of this and the two earlier spectra , reveals that all the main absorption and emission peaks have shifted to the red by 100 - 200 , depending on the wavelength . a similar , but smaller shift also occurred between nov 29 and dec 5 . the immediate interpretation is that we are observing the effect of the inward motion of the photosphere through the ejecta as they expand and thin out . note that this recession takes place essentially in the lagrangian mass / velocity frame , and not in the observer s ( radius ) frame , much like in the plateau phase of some sne ii . the si ii velocity measured from this spectrum is in fact only 8000 km s@xmath0 . the change in line absorption velocity is larger between dec 5 and dec 17 than it was between nov 29 and dec 5 because the time interval between the spectra is larger . all absorptions move by about the same amount , and the emissions move as well . this is obvious given our explanation for what the emissions really are : if the blanketed regions move , the opacity windows must move along with them , just like in p - cygni profiles caused by a single line , a situation of which the case of sn 1997ef represents after all just an extreme extension . looking in more detail , the shoulder which affected the si ii line near 5900 on dec 5 seems to have disappeared on dec 17 , leaving serious questions about its identity . the absorption near 5000 now shows more structure , with minima at 4800 and 4950 and two shoulders at 5100 and 5300 , which resembles quite closely the typical profile observed in sne ia near maximum and soon thereafter . in sne ia the entire feature is attributed mostly to fe ii lines , and this was confirmed in sn 1997ef on both nov 29 and dec 5 . the synthetic spectrum computed with co100 is shown as the dashed line in figure 4 . the structure in the fe ii absorption is reproduced reasonably well , and the width problem now does not affect either the feature at 4300 or that near 7500 . this is probably because the outer high velocity regions are so far above the photosphere now that they do not really cause any significant absorption , and so the original model co100 gives an acceptable description of the density structure in the intermediate velocity regions . unfortunately this spectrum does not cover the ca ii ir triplet , which is predicted not to blend with o i 7774 in the co100 model . the synthetic spectrum was computed with the following parameters : @xmath67 days , @xmath68 ( erg s@xmath0 ) , @xmath69 km s@xmath0 . the temperatures are lower than on dec 5 but comparable to those on nov 29 , and @xmath27 is actually larger than on both previous epochs , although the mass above the photosphere is now as large as 5m@xmath2 . compared to the dec 5 model , the abundance of o is reduced and that of s increased , which is the behaviour expected as deeper and deeper regions of the ejecta are probed . the reduced velocity is the reason for the de - blending of many features . there are two major shortcomings in the model : 1 ) the position of the peak at 5500 is not correctly reproduced . this is clearly the consequence of the absorption at 5400 not being correctly reproduced , so that the ` optical depth window ' starts too far to the blue . this feature is not well reproduced in the spectra of sne ia either , so it is quite likely that some fe - group lines are missing or their strength is not correct . 2 ) the si iiline is too narrow , or rather the very triangular shape of the blue side of the absorption can not be reproduced . only another si ii line is active in that region ( si ii 5979 , observed near 5800 ) , but it is too weak . in any case it would be difficult to explain the observed profile with just one line , so we really have no explanation for what might affect the spectrum . we have also modelled the spectrum using our modified density distribution ( figure 4 , continuous line ) . as we could expect , since the near - photospheric region is not affected by the modification , at this epoch the two synthetic spectra are essentially identical . the spectrum computed with the modified outer density profile has @xmath70 and @xmath71 . the bolometric correction is still large , @xmath72 , and @xmath73 . it is very interesting that the model luminosity , as obtained from fitting the spectrum , is actually larger on dec 17 than on dec 5 . the bolometric light curve appears then quite different from both the @xmath13 and the @xmath74 curves . the large bolometric correction is now caused mostly by the red part of the spectrum , with the synthetic @xmath75 band still rising so that the bolometric light curve rises slowly to a rather delayed maximum . near - ir photometry would have been useful to confirm this prediction . the 24 dec spectrum is the first of a set of three spectra modelled in this paper which extend the time coverage beyond the near - maximum epochs considered also in paper i. the 24 dec spectrum is shown in figure 5 . it does not extend far to the red , but in the blue it has a good s / n down to 3700 , thus covering the ca ii h&k region . the epoch of the spectrum is 35 days after explosion , for which we chose nov 20 . the spectrum has @xmath76 , a drop of 0.2 mag in only one week from dec 17 , and lies therefore well in the declining part of the @xmath13 light curve . the si ii line is now quite narrow , and its core has a velocity of 5400 km s@xmath0 . this and all other absorptions continue to move to the red . following the trend already noticeable on dec 17 , many small features which earlier blended into the broad troughs are beginning to be resolved , marking the progressive inward motion of the photospere . the broad absorption near 6000 is the region that changed the most with respect to dec 17 . in particular , a separate absorption is now visible at 5700 . we modelled the spectrum using model co100 with the outer modification introduced as discussed above . as usual , @xmath10 and @xmath25 were selected so as to get as good a fit as possible to the observations . the input parameters for the best model , which is shown as the continuous line in figure 5 , were @xmath77 days , @xmath40 ( erg s@xmath0 ) and @xmath78 km s@xmath0 . note that @xmath10 has the same value as on nov 29 , indicating that the bolometric light curve is now declining . the photospheric velocity is now very low , as is also indicated by the si ii line velocity , and is actually slightly below the inner edge of the density distribution of model co100 ( cf . paper i , figure 3 ) . therefore , in order to produce a reasonable spectrum we had to introduce a second change to the co100 density structure , extending it down to @xmath79 . we used @xmath80 for this inner extension . this is only a first approximation , but it is obvious that if there is a photosphere there must be opacity right down to it , and this is not compatible with the ` density hole ' predicted by model co100 below 5300 km s@xmath0 . since the photosphere is so deep , the inclusion of the flat outer density profile derived from the earlier spectra has no effect on the synthetic spectrum . even when the flat outer density is included , o i 7774 and the ca ii ir triplet are predicted to be separated in wavelength . this region is not covered on dec 24 , but our prediction will be confirmed in the jan 1 spectrum ( see next section ) . the small inner extension to the density does not affect the total mass much , so that at this epoch the total mass above the photosphere is 8.4m@xmath2 . a major change in this model is in the abundances . as shown in figure 7 of paper i , the oxygen - dominated envelope extends down to 6500 km s@xmath0 . on dec 24 the photosphere is well below this point , and si , s and the fe - group have larger abundances . we fitted the abundances to obtain a best fit , and found that s is the element whose abundance must increase most , at the expense of o. the behaviour of the abundances as determined from spectrum synthesis is reviewed in the discussion . most line identifications are the same as on earlier epochs . the absorption at 5700 is due to na i d . overall , the synthetic spectrum is a good match to the observed one . the major shortcoming is that the emission peak near 6500 is not reproduced . net emission may be present there , although the wavelength does not match that of the expected line of [ o i ] 6300 . there is a wavelength coincidence with h@xmath81 , but this is not expected to be a strong feature in a sn ic ejecta . the synthetic spectrum has @xmath82 and @xmath83 . the bolometric correction is still large , @xmath84 , so that @xmath52 . therefore , it appears that bolometric maximum was reached near dec 24 , much later than @xmath13 or @xmath74 maximum . a lot of the flux is still released in the near - ir , in particular in the ca ii ir triplet peak . the next spectrum in our series is rather close in epoch to the previous one but , with an epoch of 42 days , it captures a moment when the light curve begins to settle on the tail . although the s / n is only moderately high , the ample wavelength coverage ( 3400 to 8900 ) makes it an interesting spectrum to model . the spectrum is shown in figure 6 . comparing it to the dec 24 spectrum , several signs of development can easily be noticed : 1 . the ratio of the three peaks in the blue , at 4000 , 4500 and 5500 is much steeper , indicating a shift of the flux towards the red which clearly must be the consequence of a reduced temperature . the absorption complex near 6000 has now split into two components . one is centred near 5700 and is due to na i d , which is now sufficiently strong and isolated because of the reduction in strength of neighbouring lines coming from more highly excited levels that it gives rise to its own p - cygni emission , near 5900 . to the red of na i d the si ii line is still strong , near 6250 , and is accompanied by a weaker absorption near 6100 , which was also present on dec 24 . two absorptions are developing at about 6800 and 7100 . the o i 7774 line and the ca ii ir triplet are now well separated , as was already predicted by our synthetic spectrum on dec 24 . the o i line is rather narrow , while the ca ii absorption is broader . this is a consequence of the different run of excitation with radius ( excitation falls more steeply for o than for ca , and so does the line optical depth ) . 5 . emission in the ca ii ir triplet is now very strong , much stronger than the corresponding absorption : this indicates that net emission is beginning to be an important contribution to spectrum formation . the line blueshift is further reduced : the si ii line now has a measured @xmath85 km s@xmath0 . this is now significantly smaller than than the inner boundary of the original co100 density distribution . since the essentially photospheric nature of the spectrum is proved by the progressive evolution of all features , a viable model must include a low velocity extension to the density profile . therefore , we modelled the spectrum using as a starting point the co100 density distribution modified at high velocity as described earlier . this modification has no effect on the emerging spectrum at these later epochs : the maximum observable matter velocity , measured at the blue edge of the ca ii ir triplet is only 23000 km s@xmath0 . we extended the density distribution inwards to @xmath86 km s@xmath0 , and tried various density laws below 5000 km s@xmath0 . our first conclusion is that the density must be increasing inwards : if we use a decreasing density a shell forms near 5000 km s@xmath0which affects the lines ; if the density is constant below 5000 km s@xmath0the value near the photosphere is too low and the lines are too weak . on the other hand , if the density increases too steeply the synthetic lines tend to be too sharp . as a best approximate solution we adopted a power law with @xmath80 . our final model has parameters @xmath87 days , @xmath88 ( erg s@xmath0 ) , @xmath89 km s@xmath0 , which is close to the observed si ii velocity . it is interesting that the simultaneous reduction in both @xmath10 and @xmath25 means that the electron temperature @xmath90 ranges between 4500 and 6500k , which is close to the values it had on previous epochs . this explains why the naure of most line features is unchanged throughout the evolution of the sn . the synthetic spectrum is shown as the thin line in figure 6 . overall , the quality of the fit is rather good , confirming the correctness of our assumption . major shortcomings are : 1 ) the strength of the peak at 4500 . this is probably due to having somewhat too much opacity in the uv . 2 ) the strength and position of the peak near 5500 . this was a problem also on dec 24 and we explained it as the consequence of missing line strength near 5400 . 3 ) flux is missing in the ca ii ir triplet emission . this is because net emission - following collisional excitation - is not included in our code . the difference between the observed and synthetic peak is to be ascribed to net emission , which is responsible for about half of the total line flux . other parts of the spectrum are very well reproduced , e.g. the o i and ca ii ir triplet absorption , the structure of the si ii line region , where the weak absorption at 6000 is attributed to fe ii 6148 and the emission near 6500 is now correctly reproduced , confirming that the peak on 24 dec was probably not [ o i ] 6300 . the weak absorptions at 6800 and 7100correspond to lines of o i 7001 and to several fe ii lines , respectively , but the synthetic lines are too weak . fe ii lines are becoming stronger as the abundance of fe - group elements is higher , while that of o is further decreased . the inward extension of the density affects the ejecta mass , adding about 1.5m@xmath2 between 3000 and 5000 km s@xmath0 , but it only has a small effect on the kinetic energy , adding only @xmath91erg in that velocity shell . the synthetic spectrum has @xmath92 , which compares well with the observed @xmath93 . synthetic @xmath94 , showing that the spectrum is rather red . the @xmath75 magnitude drops less than both @xmath74 and @xmath13 , and the bolometric correction is still large , @xmath95 , so that @xmath96 , but this neglects the net emission in the ca ii ir triplet , so the actual value is probably smaller . this is the last spectrum available , corresponding to an epoch well on the light curve tail , with @xmath97 . the spectrum , shown in figure 7 , shows a clear evolution from jan 1 . in particular , the flux peak has moved to the red quite noticeably , indicating a temperature drop . looking at the spectrum in detail , one can notice that the features in the blue have changed little , and so has the region furthest to the red , although the ca ii ir triplet emission is now stronger . on the other hand , the region around 6000 has changed quite significantly . the flux peak has moved from 5500 to 6000 , the absorptions at 5700 ( na i d ) and at 6200 are much stronger . the si iiline has essentially disappeared , all that is left being a weak feature near 6400 , which is too red to be identified with the si ii doublet . in fact , a si ii velocity can not be measured at this epoch . still , many features have persisted from earlier epochs , and these have shifted further to the red , indicating the presence of significant amounts of material at very low velocity in the ejecta . the overall nature of the spectrum is still photospheric , but the incidence of net emission is higher than on jan 1 , which will limit our ability to fit the spectrum using the mc code . we used the density distribution discussed in the previous section , including both the outer @xmath1 part above 25000 km s@xmath0 , which has no effect on the synthetic spectrum , and the inner @xmath98 extension between 3000 and 5000 km s@xmath0 , but we had to extend the density distribution further inwards , so that we could obtain an appropriate temperature and spectrum . in keeping with the hydrodynamical prediction of a ` density hole ' in the centre of the ejecta , we tried to use the largest possible value of the density power law index allowed by the quality of the synthetic spectrum . we selected a model based on the overall fit , but we could not reproduce the observations as well as on previous epochs . we could accomodate an innermost density law @xmath99 below 3000 km s@xmath0 , which is an attempt to include an inner ` hole ' in density . this extension adds a further 0.5m@xmath2 to the ejecta mass , but only less than @xmath100erg to the explosion kinetic energy . our best fit spectrum , shown in figure 7 , has parameters @xmath101 days , @xmath102 ( erg s@xmath0 ) and @xmath103 km s@xmath0 . the temperature is significantly lower than on jan 1 , with @xmath104 ranging from 6000 to 4000k . because of the rather flat density structure at low velocity , the temperature is almost constant ( @xmath105k ) at @xmath106 km s@xmath0 , which is where the spectrum is formed . this helps to keep the lines sufficiently broad . the abundances are similar to those of jan 1 , but oxygen is further reduced and s is increased , reflecting the expected trend as smaller and smaller velocities are sampled . admittedly , the synthetic spectrum is not a very good fit , but almost all the observed features are at least reproduced . the two emission peaks at 4600 and 5500 are much too strong in the model , which affects the synthetic @xmath13 magnitude . the observed value is @xmath107 , and the synthetic spectrum has @xmath108 . the @xmath74 flux is also somewhat overestimated at @xmath109 . possible reasons for the excessive strength of the synthetic peaks at 4600 and 5400were already given in the previous section . in the @xmath13 region , an absorption near 5600 is also not reproduced . a similar feature was also observed on jan 1 , near 5650 , but it was weaker then , while it was absent on dec 24 . the absorption might be attributed to he i 5876 at a velocity of about 13000 km s@xmath0 . he may be expected to be found in the ejecta as a leftover of the star s he - shell . if he is distributed in a shell , he atoms may only be excited at a rather advanced epoch , when the @xmath110-rays and positrons from the decay of @xmath5co can penetrate the ejecta and reach the shell . however , the velocity implied by the position of the absorption is low for a hypotetical he shell , and the apparent shift of the feature between jan 1 and jan 26 does not support this scenario either . alternatively , he can be produced with 56ni in the deepest layer as the result of alpha - rich freezeout ( see paper i , figures 5 and 6 ) . in this case he would be found together with co / fe , so the he velocity could take any value depending on the 56ni mixing . however , in this case non - thermal excitation of he would take place immediately , and we should see the he i line as soon as the photosphere reaches the layer where he is present . if he is located at about 14000 km s@xmath0 , we should have seen the he iline as early as dec 5 . the above problems notwithstanding , as long as no other candidate identification is available he must be regarded as at least a possibility . observatons of he i 10830 would certainly help settle this issue . in any case , given the moderate strength of the line , the he mass involved may not have to be large . the model predicts a moderately strong absorption at 6300 , attributing it to a blend of the si ii doublet and , mostly , of fe ii 6417 , 6456 . this feature is present also in the observed spectrum , but it is much weaker . the reason for the discrepancy is clearly the emission at 6300 , which must signal the onset of the corresponding [ o i ] line . another net emission , which is not reproduced by the model , is visible near 7300 . this is clearly ca ii ] 7292 , 7324 . together with the ca ii ir triplet , these are the only three net emissions in the optical spectrum . this is a typical situation for sne ic at these epochs ( cf . sn 1987 m at 60 days after maximum , swartz et al . 1993 ) . comparing the spectra of sn 1997ef and sn 1987 m , apart from the obviously broader lines in sn 1997ef , it is possible to note that in sn 1987 m the fe iilines are weaker and that the candidate he i line is not seen . sn 1987 m is thought to have produced roughly as much @xmath5ni as sn 1997ef ( 0.15m@xmath2 ; nomoto , filippenko , & shigeyama 1990 , paper i ) , but this could be an overestimation . classical , low energy sne ic are not expected to eject any he , but this might not be the case for sn 1997ef , and it certainly is not for the other hypernova , sn 1998bw , where he i 10830 is observed ( patat et al . 2000 ) . on the other hand , at a comparable epoch the [ o i ] 6300 emission is much stronger in sn 1998bw than in sn 1997ef , supporting the idea that sn 1998bw came from a more massive progenitor . further detailed study of the he ( and o ) content of hypernovae would certainly be worthwhile . the synthetic spectrum has a small @xmath65 ( 0.01 ) , and @xmath111 . the actual value of @xmath65 may be even smaller , and possibly negative , since the model @xmath13 flux is certainly overestimated by about 0.15 mag , and the net emission in the ca ii ir triplet is not included in our estimate . therefore , the value of @xmath53 has an uncertainty of at least 0.2 mag . as we have discussed above at great length , modelling the time evolution of the spectrum of sn 1997ef has revealed two main inconsistencies with the explosion model co100 , which we used in paper i to reproduce both the light curve and the near - maximum spectra . the basic parameters of the observed and synthetic spectra are recapped in table 1 . firstly , the high line velocities observed near maximum in several strong lines , notably the ca ii ir triplet and the si ii doublet , could not be reproduced with the rather steep density profile of co100 , but require a flatter outer density law , @xmath1 , at @xmath21 km s@xmath0 . secondly , the fact that the january spectra are still predominantly photospheric in nature , showing low line velocities ( @xmath112 km s@xmath0 ) is not compatible with the presence in co100 of an inner ` density hole ' . this ` hole ' was the result of depositing the kinetic energy at a radius in the progenitor which results in the ejection of exactly the amount of @xmath5ni necessary to reproduce the sn tail luminosity . matter below that radius , known as the ` mass cut ' , is assumed to fall back onto the compact remnant . clearly , material must exist at low velocities , and we found that an inner extension below 5000 km s@xmath0with a @xmath80 density law down to 3000 km s@xmath0 and a @xmath113 law below that , to reproduce the inner ` density hole ' , is the best overall solution , although we still can not get perfect fits of those very late epochs , even ignoring the net emission which is clearly present in some lines . in figure 8 we show the density profile of the original co100 model and our modified one . the outer density extension adds significanlty to the kinetic energy ( @xmath114erg ) , but only marginally to the ejecta mass ( 0.25m@xmath2 ) , while the inner extension adds about 1.65m@xmath2 to @xmath115 but has a negligible effect ( @xmath116erg ) on the explosion kinetic energy . our modified model has a total @xmath117m@xmath2 and @xmath118erg , compared to @xmath119m@xmath2 and @xmath120erg for co100 . these new values reinforce the classification of sn 1997ef as a hypernova , which is based on the @xmath22 being larger than @xmath47erg . since we have a @xmath1 density law on the outside , and an essentially flat law deep inside , and since our density distribution was derived starting from that of co100 , whose relic is the steep density gradient in the intermediate part , one might argue that a @xmath1 density law might work well throughout the ejecta . we have tested such a possibility , but although it works reasonably well for the outer part , at @xmath121 km s@xmath0 , it gives rise to very sharp - lined spectra at the later epochs , when the photosphere is in the flat part of the ejecta , at @xmath122 km s@xmath0 . also , joining directly the inner flat part and the outer @xmath1 part beyond 10000 km s@xmath0 would give a sn with extremely large @xmath115 and @xmath22 , and the synthetic spectra for the epochs when the photosphere is near 10000 km s@xmath0 would have very deep lines . we have to ask ourselves the question what could give rise to the observed deviation from the hydrodynamical model . the flat outer part may be due to mass loss at high rate during the presupernova stages , or to a transition between the oxygen shell and an outer helium shell , some evidence for which may be present as a weak he i 5876 line in the two january spectra , although at a velocity ( @xmath123 km s@xmath0 ) which is smaller than what would be expected . branch ( 2000 ) offered a similar solution , and his ad hoc density law was even flatter @xmath14 than ours . as a result he offers an estimate of the kinetic energy above 7000 km s@xmath0 as @xmath17erg . our value above the same velocity is somewhat smaller but comparable , @xmath124erg . also , he suggests that the ejecta mass above 7000 km s@xmath0 is @xmath125m@xmath2 , and our value is again smaller but comparable , @xmath126m@xmath2 . the location of the inner density cutoff , on the other hand , was determined essentially by demanding that the correct mass of @xmath5ni be ejected from the progenitor . clearly , the estimate of the position of the cut - off was incorrect . although it is possible that this may be due to an incorrect calculation of the progenitor s structure , it is more tempting to attribute the problem to some asymmetry in the explosion . if the explosion was asymmetric , similar to what has been suggested for the hypernova sn 1998bw in order to explain its connection with a grb , it is quite possible that most @xmath5ni was produced near the beam axis , while away from that axis burning would be less efficient , and would terminate at intermediate elements such as si or s ( maeda et al . 2000 ) . clearly one then has to place the mass cut deeper in order to achieve the require ejected mass of @xmath5ni . this is also supported by our derived abundance distribution , which suggests that the s abundance is still increasing at the lowest velocities sampled , about 2000 km s@xmath0 . the abundance distribution as derived from our models is displayed in figure 9 , but we must keep in mind that , since we used homogenised compositions above the photosphere , all sharp composition boundaries are smoothed out . on the other hand , no clear spectral evidence - such as sudden changes in the properties of the lines - is visible for strict abundance stratification . the presence of an inner density core was also suggested in paper i to explain the observed deviation of the synthetic light curve computed with model co100 from the observations at advanced phases : the synthetic light curve has a steeper slope than the observed one , indicating that @xmath110-ray trapping is more efficient than in the model . this could be achieved if @xmath5ni was distributed deeper than it is in co100 . in order to verify our findings , we used our derived density and abundance distribution and computed a synthetic light curve in spherical symmetry . we used two different codes , one based on monte carlo methods and the other , a radiation hydrodynamics code , which was used in the light - curve computation in paper i. the monte carlo code is based on the simple code developed for sne ia described in cappellaro et al . @xmath110-ray deposition in the expanding ejecta is computed following the random walk of gamma - ray photons adopting a constant @xmath110-ray opacity @xmath127@xmath128g@xmath0 . once they deposit their energy , the @xmath110-rays are assumed to generate optical photons on the spot . the random walk of the optical photons through the ejecta is then also followed in monte carlo , assuming a constant optical opacity @xmath129 . the monte carlo scheme is able to model efficiently the random walk of photons through the ejecta - and thus to take into account the delay between the emission of a photon and its escaping from the sn nebula , which determines the initial shape of the light curve . the value of @xmath129 depends on the temperature and composition , and it is possible that it changes with time . thus we had to find a convenient approximate value for this particular model of sn 1997ef . since @xmath129 affects the light curve essentially near maximum , we found that a value @xmath130 @xmath128g@xmath0 gives a good fit to the light curve around maximum . this value is smaller than for sne ia , as might be expected since line opacity is much stronger in the fe - dominated ejecta of a sn ia . the synthetic light curve in monte carlo is compared to the one obtained with the radiation hydrodynamics code . as described in paper i , the code solves a multi - frequency radiative transfer equation in the fluid s comoving frame , using the feautrier method iteratively with an approximate lambda operator . the energy equation of gas plus radiation and the radiation momentum equation are coupled to the transfer equation in order to follow the evolution of gas temperatures . the code also uses average opacities , and the values used were the same as in the monte carlo code . the synthetic bolometric light curves are shown in figure 10 . except for a slight difference around the maximum , which is probably to be ascribed to monte carlo noise , the codes produce very similar light curves . therefore , it is safe to use the simple monte carlo code in calculating bolometric luminosities if a reasonable opacity is chosen . the synthetic light curves reproduce our derived bolometric curve reasonably well around maximum . the @xmath13 light curve can consequently be reproduced if the bolometric correction as resulting from our synthetic spectra is taken into account . in both codes a @xmath5ni mass of 0.13m@xmath2 was used . in order to achieve a rapid rise of the light curve , consistent with our dating , we had to distribute @xmath5ni homogeneously throughout the ejecta . both light curves reproduce the maximum quite well , but their decline on the tail is much too steep . the observed @xmath13 light curve after about day 60 appears to follow the @xmath5co decline rate , implying constant and complete deposition of about 0.07m@xmath2 of @xmath5ni . however , our codes predict for those advanced epochs a deposition fraction of less than 0.5 and steadily decreasing with time . in figure 11 we compare the observed velocity of the si ii doublet with the run of @xmath131 as derived from the two light curve codes and with the values used as input for the calculation of the synthetic spectra . the two codes are in good agreement with one another , although the mc values are consistently larger than those of the radiation hydro model . this is probably due to different zoning in the two codes . however , the observed points are lower than both models , and the difference increases with the epoch . the velocities used for the spectral calculations , on the other hand , start lower than the observed values but slowly approach them . this is understandable , as it can be expected that at early times the si ii line forms above the photosphere since it is very strong . later , as the line becomes weaker , it forms closer and closer to the photosphere . the fact that the value of @xmath131 tends to flatten out at advanced epochs , and does not reproduce the observations , is not new ( see , e.g. , iwamoto et al . 1998 on sn 1998bw ) , and it probably indicates that the assumption of a gray photosphere is not good at advanced epochs . however , the difference at early phases can not be due to that effect . one possibility is that we have somewhat overestimated the density at the highest velocities in our spectral calculations , where we were guided more by the ca ii ir triplet than by the si ii doublet . note that the velocity derived from model co100 in paper i was much smaller than the observed values . another possibility is that our assumption of a gray opacity again fails in those outermost layers , as shown perhaps by the fact that only the ca ii ir triplet is active there . note that all of these remarks apply to a spherically symmetric situation . another inconsistency between the light curve and spectral calculations is that if we integrate the ni mass as the sum of the abundances of fe , co and ni in the various synthetic spectra we obtain an fe - group mass of only @xmath132m@xmath2above 2000 km s@xmath0 , which is much less that what is necessary to power the light curve . it is unlikely that our spectrum synthesis may have given such a large error , especially since the fe lines are already very strong in most spectra . so we have a siuation where at early times we need more surface @xmath5ni to power the light curve than we see in the spectra , while at late times the @xmath5ni we have used to reproduce the light curve at peak does not deposit sufficiently to reproduce the light curve slope and brightness . the late light curve appears to require the complete deposition of @xmath133m@xmath2 of @xmath5ni . the two problems may be independent , but they may also be related . one possibility to explain the late light curve behaviour is that , in a spherically symmetric scenario , additional @xmath5ni ( @xmath134m@xmath2 ) could be located deep in the ejecta , below 2000 km s@xmath0 . if the density is sufficiently high there the @xmath110-rays produced by the decay of this @xmath5ni may deposit completely . the presence of very low velocity ejecta would further reduce the mass of the expected compact object , and may be in contradiction with the distribution of @xmath5ni in models of the progenitor evolution and explosion . this would not affect the situation at early times . another possibility is that the ejecta may not be spherically symmetric , in either mass or abundance distribution , or both . if some @xmath5ni is ejected at low velocities together with other elements in some direction , the @xmath110-rays produced by its decay chain could be efficiently trapped even at advanced epochs . the optical photons produced by the thermalisazion of @xmath110-rays and positrons in this hypothetical high density region would only be able to escape at advanced phases , and so they would have little effect on the early time spectra but they could power the late light curve . at the same time , some material has been ejected at high velocity , possibly in a jet - like form , to reproduce the sharp rise of the light curve by high - velocity @xmath5niand to explain the broad lines of si ii and ca ii ( @xmath135 km s@xmath0 ) at early epochs . in this scenario , the uniform mixing of @xmath5ni which was used to reproduce the sharp rise of the light curve might refer only to the @xmath5niejected in the jet , while the low velocity @xmath5ni would enter play only later . this scenario might help explaining the discrepancy between the light curve and spectroscopic @xmath5ni masses . if most of the @xmath5ni is ejected in a preferred direction ( e.g. in the jet that might be observed as a grb if the viewing angle is favourable ) , this @xmath5ni may power the early light curve by depositing its @xmath110-rays in the neighbouring ejecta , which have lower velocity , higher density and a smaller fe - group abundance . however , if we view the event from an angle sufficiently far from the jet direction , the spectrum we see would be produced in a region where the abundances are different from those found in the jet hence the smaller spectroscopic mass of @xmath5ni . this is a possibility for sn 1997ef , since there is no positive identificatin of a grb counterpart to the sn event . the possibility that the explosion was asymmetric is in line with existing models of energetic supernova explosions linked to ghrb s ( macfadyen & woosley 1999 , hflich et al . 1999 , khokhlov et al . 1999 , maeda et al . 2000 ) , and it deserved further study with multi - dimensional radiation hydrodynamics codes . we have shown in this paper how spectrum synthesis can be used not only to verify explosion models of sne , but also to improve on them . given our findings , it would now be very interesting to see hydrodynamic calculations of the explosion , in one or more dimensions , which reproduce closely the observed bolometric light curve . the derived density distribution , ejecta mass and kinetic energy may or may not be close to what we have obtained spectroscopically , but in any case such a model would have to be tested also for its ability to reproduce the sn spectral evolution . our study also shows that fitting an observed @xmath13 curve with a synthetic bolometric one may be somewhat misleading , since the bolometric correction can be significant . we have discussed in 9 the reason for the difference between the density distribution derived from spectral modelling and that obtained from 1d hydrodynamic calculations of the explosion of the stripped core of a massive star , and remarked that the presence of significant amounts of ejected matter at low velocity may be due to some asymmetry in the explosion . a similar conclusion was reached for the other type ic hypernova , sn 1998bw , but in that case it was based on somewhat different evidence , most importantly the connection with a grb ( e.g. wheeler 2000 ) . the early - time spectra of sn 1998bw have even broader and more blended lines than those of sn 1997ef , but a clear evolution towards lower line velocities and less line blending is not seen as clearly as in sn 1997ef , while the relatively early development of nebular emission is seen in both objects . this may be understandable since although the ejecta mass of sn 1998bw was larger than that of sn 1997ef , even if we use our upwards revised value for the latter , the kinetic energy of sn 1998bw appeared to be much larger ( nomoto et al . branch ( 2000 ) in a spectroscopic study similar to the one he performed for sn 1997ef , suggested an extremely large @xmath22 for sn 1998bw ( @xmath136erg above 7000 km s@xmath0 , i.e. a factor 2 larger than his value for sn 1997ef above the same velocity ) . we have performed a preliminary analysis of sn 1998bw using the technique described in this paper , and find a similarly large value ( nakamura et al . 1999 ) . if sn 1997ef was also a highly asymmetric explosion , although weaker in energy and ejecting a smaller mass than sn 1998bw , why was a grb not positively detected ? the most likely possibilities are that either the jet - axis of sn 1997ef was not oriented exacly along the line - of - sight , or that a weaker explosion energy reduced the beaming so that a grb was not formed . both scenarios might be able to explain many of the observed features of sn 1997ef , in particular the lower expansion velocities and the smaller measured @xmath22 . when we apply a spherically symmetric model , we estimate @xmath22 by integrating a 1d model around a sphere . therefore , if the sn is observed on or very close to the jet - axis , like sn 1998bw , @xmath22 could be grossly overestimated hflich et al . 1999 ) . on the other hand , the difference between the velocity of the spectral lines in the two sne is real , and so is the difference in luminosity at late times . that sn 1998bw produced nore @xmath5ni is also confirmed by the estimate for the @xmath5ni mass ( @xmath137m@xmath2 ) obtained from the nebular lines , which does not depend much on the asymmetry ( see danziger et al . 1999 ; nomoto et al . 2000 ) . therefore , even though inclination may be a factor , that there is some intrinsic difference between the two objects appears unavoidable , at least as long as the relative distance estimate is reliable . finally , we note that another sn 1997ef - like type ic hypernova was recently observed , sn 1998ey , whose spectra appear to be identical to those of sn 1997ef , and for which there is also no observed grb counterpart ( garnavich et al . this coincidence is striking , and calls for the accumulation of more data on type ic hypernova candidates . * acknowledgements * this work has been supported in part by the grant - in - aid for scientific research ( 12640233 , 12740122 ) and coe research ( 07ce2002 ) of the ministry of education , science , culture and sports in japan . it is a pleasure to thank p.garnavich and j.danziger for useful communications and the referee , d.branch , for constructive remarks . nomoto , k. , mazzali , p. , nakamura , t. , iwamoto , k. , maeda , k. , suzuki , t. , turatto , m. , danziger , i.j . , & patat , f. 2000 , in ` supernovae and gamma - ray bursts ' , ed . m. livio et al . , cambridge university press , cambridge , in press ( astro - ph/0003077 ) rrccccccccccccc 29 nov & 9 & 42.13 & 15500 & 19072 & -12.65 & 1.05 & 6123 & 7725 & 17.45 & 16.74 & 16.7 & 0.29 & -16.600 5 dec & 15 & 42.15 & 9500 & 13962 & -12.30 & 3.38 & 6128 & 9447 & 17.35 & 16.58 & 16.5 & 0.40 & -16.650 17 dec & 27 & 42.20 & 7500 & 8011 & -12.67 & 5.16 & 5291 & 6648 & 17.72 & 16.65 & 16.6 & 0.20 & -16.775 24 dec & 34 & 42.13 & 4900 & 5378 & -12.71 & 7.56 & 5602 & 7602 & 17.83 & 16.71 & 16.8 & 0.32 & -16.500 1 jan & 42 & 42.00 & 3600 & 3775 & -12.78 & 8.48 & 5426 & 6830 & 18.41 & 17.16 & 17.3 & 0.22 & -16.250 26 jan & 67 & 41.80 & 1950 & & -13.51 & 8.97 & 5232 & 5701 & 19.06 & 17.85 & 18.0 & 0.01 & -15.775	the properties of the bright and energetic type ic sn 1997ef are investigated using a monte carlo spectrum synthesis code . analysis of the earliest spectra is used to determine the time of outburst . the changing features of the spectrum and the light curve are used to probe the ejecta and to determine their composition , verifying the results of explosion calculations . since synthetic spectra computed using our best explosion model co100 are only moderately good reproductions of the observations , the inverse approach is adopted , and a density structure is derived by demanding that it gives the best possible fit to the observed spectrum at every epoch analysed . it is found that the density structure of model co100 is adequate at intermediate velocities ( 500025000 km s@xmath0 ) , but that a slower density decline ( @xmath1 ) is required to obtain the extensive line blending at high velocities ( 2500050000 km s@xmath0 ) which is the characterising feature of this and other energetic type ic supernovae . also , the inner ` hole ' in the density predicted by the model is found not to be compatible with the observed evolution of the spectrum , which reaches very low photospheric velocities at epochs of about 2 months . the ` best fit ' density distribution results in somewhat different parameters for the sn , namely an ejecta mass of 9.6m@xmath2 ( v. 7.6m@xmath2 in co100 ) and an explosion kinetic energy of @xmath3erg ( v. @xmath4erg in co100 ) . this revised value of the kinetic energy brings sn 1997ef closer to the value for the ` prototypical ' type ic ` hypernova ' sn 1998bw . the abundance distribution of model co100 is found to hold well . the modified density structure is used to compute a synthetic light curve , which is found to agree very well with the observed bolometric light curve around maximum . the amount of radioactive @xmath5ni produced by the sn is confirmed at 0.13m@xmath2 . in the context of an axisymmetric explosion , a somewhat smaller kinetic energy than that of sn 1998bw may have resulted from the non alignment of the symmetry axis of the sn and the line of sight . this might also explain the lack of evidence for a gamma ray burst correlated with sn 1997ef .
dynamical evolution of globular clusters is strongly affected by gravitational tidal shocks . when the clusters cross the disk of the galaxy they experience disk shocking ; when the clusters pass near the galactic center , they experience bulge shocking . the effects of the tidal shocks depend on the density of the background stars and are especially pronounced in the inner regions of the galaxy . tidal shocks increase the energy of random motion of stars , reduce the binding energy of the cluster , accelerate core collapse , and lead to the faster overall evolution and destruction of globular clusters ( for example , spitzer 1987 ; weinberg 1994 ; murali & weinberg 1997a , b , c ; gnedin & ostriker 1997a ) . we calculate the rate of destruction of globular clusters as a result of various physical processes : two - body relaxation , evaporation of stars through the tidal boundary , disk shocking and bulge shocking ( gnedin & ostriker 1997a ) . we use a fokker - planck code which includes tidal shocks semi - analytically . we introduce the adiabatic corrections that account for the conservation of adiabatic invariants of the fast moving stars ( gnedin & ostriker 1997b ) . these corrections reduce the energy input due to the shocks in the inner parts of the cluster . the results ( figure 1 ) show that tidal shocks dominate cluster evolution near the galactic center . overall , as many as 50% to 90% of the present sample of globular clusters may be destroyed within the next hubble time . tidal shocks destroy most easily the low - mass , low - density clusters . figure 2 shows the distribution of the inner and outer galactic globular clusters . as expected , there are no low - density clusters in the inner part of the galaxy where the tidal shocks operate most efficiently . removal of those low - mass clusters makes the mean , or the peak , of the luminosity function ( lf ) to shift towards bighter magnitudes . an apparent correlation of the masses and densities of outer clusters allows us to construct an intrinsic distribution of globulars , unaffected by the shocks . by applying the dynamical calculations , we can estimate the amount of brightening of the peak of the lf . assuming that in all galaxies the initial distribution is the same , we can reconstruct the shock history in an external galaxy and infer the peak of the original distribution of globular clusters . comparing that peak with the center of the intrinsic distribution in the galaxy , we obtain a distance estimate to the galaxy . this method is fully independent and makes unnecessary the common assumption that the peak of the lf is a standard candle . applied to the best known samples of m31 and m87 ( ostriker & gnedin 1997 ) , our method gives a distance estimate in very close agreement with that obtained with cepheids and other methods . tidal heating can be simply parametrized to study semi - analytically the evolution of the individual clusters . we derive analytic equations for the first and second order energy changes , @xmath0 and @xmath1 , of stars in the cluster ( gnedin , hernquist & ostriker 1997 ) . these equations are supplemented by the adiabatic corrections depending on the effective duration of the shock . heating on the nearly circular orbits is strongly suppressed because the `` shock '' becomes very slow ( figure 3 ) . the analytical estimates are tested against the self - consistent n - body simulations of the shocking event along a true trajectory of the cluster in the galaxy . we find a remarkable agreement with the simulations . gnedin , o. y. , hernquist , l. & ostriker , j. p. 1997 , , submitted gnedin , o. y. & ostriker , j. p. 1997a , , 474 , 223 gnedin , o. y. & ostriker , j. p. 1997b , , submitted murali , c. & weinberg , m. d. , 1997a , , 288 , 749 murali , c. & weinberg , m. d. , 1997b , , 288 , 767 murali , c. & weinberg , m. d. , 1997c , , 291 , 717 ostriker , j. p. & gnedin , o. y. 1997 , , 487 , 667 spitzer , l. 1987 , dynamical evolution of globular clusters ( princeton : + princeton university press ) weinberg , m. d. 1994 , , 108 , 1414	the semi - analytic theory of tidal shocks proves to be a powerful tool to study tidal interactions of star clusters and satellite galaxies with their massive hosts . new models of the globular cluster evolution employ a combination of analytic estimates , solutions of the fokker - planck equation and direct n - body simulations . the models predict large destruction rates for the galactic globular clusters . those on the highly eccentric orbits around the galactic center are much more likely to be disrupted than the ones on nearly circular orbits . the destruction rates are largely increased near the bulge . disruption of the low - mass clusters changes the luminosity function of the globular cluster system , shifting the peak of the luminosity function to the brighter end .
chromosomes , the genetic material of a cell , are duplicated and properly segregated before the cell divides into two daughter cells @xcite . each of the sister chromatids , that result from chromosome replication , is bound to a proteinous structure , called kinetochore ( kt ) that , in turn , is coupled to the plus ends of stiff polar filaments called microtubules ( mt ) @xcite . the negative ends of these mts are located at the poles of the fusiform structure , called mitotic spindle @xcite . the process of chromosome segregation , called mitosis @xcite , is carried out in eukaryotic cells by the dynamic mitotic spindle which is self - organized from its components for this purpose . there are strong indications that molecular motors @xcite are located at the kt - mt interface ( though not in all eukaryotes ) ; these motor proteins , which are capable of generating force by hydrolyzing atp , are believed to generate poleward force or anti - poleward force depending on the family to which they belong . however , polymerizing and depolymerizing mts are also found to make significant contributions to the forces that cause chromosomal movements @xcite . the force - generation capability of depolymerizing mts have been demonstrated _ in - vitro _ however , how a mt maintains contact with the kinetochore while depolymerizing from its tip and how it exerts force on the kt remain challenging open questions in spite of the recent progress in identifying the molecular components of a kt and their spatial organization @xcite , moreover , in most of the eukaryotic organisms multiple mts couple to each kt ; how the ( de-)polymerization of mutiple mts are coordinated is still a mystery . _ budding yeast _ is a much simpler system because each kt is coupled to only a single mt . but , even in budding yeast , the two embraced sister chromatids are coupled to two mts approaching them from the two opposite poles of the spindle . in this paper we theoretically study an even simpler system , namely an isolated kt coupled to a single mt ; this system mimics an _ in - vitro _ experiment @xcite where a single kt - mt attachment was reconstituted using purified extracts from budding yeast . in those _ in - vitro _ experiments the average lifetime was found to vary nonmonotonically with the externally applied load tension . in other words , the kt - mt attachment is stabilized ( instead of getting destabilized ) at sufficiently small load tension . similar tension - induced stabilisation of attachment between e - coli fimbriae and target cells have been interpreted in the past in terms of the concept of `` catch - bond '' @xcite . akiyoshi et al . @xcite developed a 2-state kinetic model that involves four phenomenological rate constants . by assuming the rate constants to vary exponentially with the load tension , they obtained a reasonably good fit of the model predictions with their experimental data . here we do not attempt any quantitative comparison with the experimental data reported in ref.@xcite . instead , our approach allows to derive _ analytical _ expressions for the lifetimes of the kt - mt attachments within the framework of the simple theoretical models that we develop here . these models describe the polymerization- depolymerization of the mt explicitly and capture the effects of its interaction with the coupler by a potential energy landscape that gets modified when the mt is subjected to an external tension . therefore , our results for the mean lifetimes provide deep physical insight into the interplay of opposing forces and competing kinetic processes that , together , determine the stability of the kt - mt coupler . we also make theoretical predictions on the probability distribution of the lifetimes which also indicate the probability of survival of an attachment at least for a time interval @xmath0 after the attachment is established . in principle , our new predictions can be tested by analyzing the data collected in the _ in - vitro _ experiments provided the experiment is repeated sufficiently large number of times . this paper is organized as follows . in sections [ sec - minimalcont ] and [ sec - minimaldisc ] we develop a _ minimal theoretical model _ of a device that couples a single mt with a single kt and subject it to an external tension . space is represented by a continuous variable in section [ sec - minimalcont ] and by a discrete variable in section [ sec - minimaldisc ] . our results on the distribution of the lifetimes of the model kt - mt attachments , derived analytically in section [ sec - minimalcont ] , are compared with the corresponding numerical data , obtained by numerical simulations , in section [ sec - minimaldisc ] . next , in section [ sec - realistic ] , we extend the minimal model by representing the kt - mt interaction by a more realistic potential . calculation of the full distribution of the lifetimes in this case is too difficult to be carried out analytically . therefore , in this case we have directly calculated the mean lifetime of the kt - mt attachments and found that the realistic potential leads to quantitative changes without affecting the qualitative features of the kt - mt lifetimes observed in the minimal model . finally , motivated by recent evidence that the kt - mt coupler in mammalian cells might be a hybrid nano - device , composed of spatially separated active and passive components , we have extended our study also to a simple model of a hybrid coupler in appendix [ sec - hybrid ] . in the first version of the minimal model space is represented by a continuous variable and the kinetics of the system is formulated in terms of a fokker - planck equation . we find it more convenient to derive our analytical results using this formulation . for the convenience of numerical simulations , in the next section , we discretize the same model following prescriptions proposed earlier by wang , peskin and elston ( wpe ) @xcite . we also compare the results of the two versions . for the minimal model we do not need to make any assumption about the molecular constituents and their spatial organization in the coupler . the only postulates are as follows : + _ postulate ( a ) _ : the energy of the system is lowered monotonically with the increasing overlap between the inner surface of the cylindrically shaped coupler and the outer surface of the mt , and the corresponding binding energy is proportional to the projected length of this overlap along the mt axis ; + _ postulate ( b ) _ : the rate of depolymerization of a mt is suppressed by external force . to our knowledge , the postulate ( a ) is consistent with the scenario envisaged originally by hill @xcite , except for the fact that the `` roughness '' of the mt - coupler interface in the hill model , which is also an integral part of our realistic model ( described in the next section ) , is not captured by the minimal version of our model . the existence of the sleeve was postulated by hill @xcite long before information on the the molecular structure of mt - kinetochore coupler began to emerge from experiments . the current knowledge on the inventory of the mitotic machinery have identified the plausible candidates that give rise to the effectively sleeve - like coupler . the dam1 ring ( also called dash ) and the ndc80 complex seem to be the strongest candidates for the components of the coupler @xcite . the postulate ( b ) is supported by the observation of franck et al . @xcite in their _ in - vitro _ experiments that the rate of depolymerization of mt is suppressed by externally applied tension . it has been known from quite some time that the tips of the depolymerizing protofilaments are curled radially outward from the central axis of the cylindrically shaped mt . based on this fact , franck et al.@xcite speculated that the dam1 ring complex may transmit the external tension to the curled tips of the protofilaments thereby tending to straighten them . such straightening of the protofilaments is likely to suppress the tendency of the protofilaments to peel away from the depolymerizing tip of the mt . it is worth emphasizing that our results presented here require only the validity of the postulates ( a ) and ( b ) irrespective of the nature of the underlying cause of their validity . the length of the `` coupler''is denoted by @xmath1 . in the continuum formulation of the stochastic kinetics , the time - dependent variable @xmath2 denotes the instantaneous length of overlap between the outer surface of the mt and inner surface of the coaxial cylindrical coupler . thus , @xmath3 and @xmath4 corresponds to minimum and maximum overlap , respectively ; the mt is just on the verge of exiting the coupler when @xmath3 . guided by our postulate ( a ) , in the minimal version of our model we capture the mt - coupler interaction by a potential energy that is proportional to the overlap @xmath5 , i.e. , @xmath6 parametrized by @xmath7 , the depth of the potential at @xmath4 . the form of this interaction in the more realistic version of our model is given in section [ sec - realistic ] . similarly if the mt is pulled outward away from the coupler by a constant external force @xmath8 ( also referred to as a load tension ) , then this force can be derived from the corresponding potential , @xmath9 @xmath10 and @xmath11 are plotted in fig . ( [ potential ] ) . net potential felt by the kinetochore is @xmath12 and @xmath11 are plotted against the instantaneous length of overlap @xmath2 between the outer surface of the mt and inner surface of the coaxial cylindrical coupler . ] we use the symbols @xmath13 and @xmath14 to denote the addition and removal rates of the tubulin subunit from the mt . when subjected to a load tension , according to our postulate ( b ) , the depolymerization rate @xmath15 of the mt decreases with increasing strength of the tension @xmath8 @xcite . more specifically , we assume an exponential decrease of @xmath15 such that @xmath16 where @xmath17 is the depolymerization rate in the absence of any external force and @xmath18 is the characteristic load force at which the mt depolymerization rate is an exponentially small fraction of ( i.e. , a factor of @xmath19 smaller than ) @xmath20 . our motivation for assuming the specific exponential form ( [ eq - fdepbeta ] ) and the key role of the characteristic force @xmath18 will be discussed later in this paper . stochastic movement of the system can be described by a _ hypothetical diffusing particle in an external potential u(x)_. the drift caused by the addition and removal of the tubulin from mt give rise to an additional term @xmath21 to the drift velocity . note that the position of the hypothetical particle ( i.e. , the overlap @xmath5 ) can change by any of the following processes : + ( i ) even in the absence of any external force , the mt has a natural tendency to approach @xmath4 because the system can lower its energy by increasing the overlap between the outer surface of the mt and the inner surface of the coupler . in other words , the potential @xmath10 gives rise an effective force that spontaneously pulls the mt into the coupler . + ( ii ) the external load tension can pull the mt outward with respect to the coupler thereby decreasing the overlap @xmath5 . alternatively , an external force can push the mt into the coupler as long as @xmath5 remains non - zero . + ( iii ) the overlap @xmath5 can decrease , as long as it is non - zero , because of the depolymerization of the mt . + ( iv ) the overlap @xmath5 can increase because of the polymerization of mt . + ( v ) finally , even in the absence of polymerization or depolymerization and force - induced movements , the position @xmath5 can change because of its one - dimensional diffusion ; @xmath22 being the corresponding diffusion constant . therefore , in the overdamped limit , @xmath23 where @xmath24 is the phenomenological coefficient of the viscous drag and @xmath25 is a gaussian white noise . let @xmath26 denote the probability of finding the overlap @xmath5 at time @xmath0 . the fokker planck equation @xcite , @xmath27 that corresponds to the stochastic differential equation ( [ eq - langevin ] ) , is essentially a diffusion equation with an additional drift term where the time - independent , but tension - dependent , net drift velocity is @xmath28 and d is the diffusion constant . the interplay of the four key components are shown clearly in eqn.([myc ] ) . the external load tension @xmath8 competes against @xmath29 while mt depolymerization competes against polymerization . the load tension @xmath8 has two mutually opposite effects on the kinetics : ( i ) it directly tends to decrease @xmath5 by pulling out the mt from the coupler , and ( ii ) it indirectly assists increase of @xmath5 by suppressing @xmath15 that competes against @xmath13 . the equation ( [ main ] ) can also be written as an equation of continuity @xmath30 for the probability density @xmath26 , with the probability current density @xmath31 \label{eq - probcurr}\ ] ] the expression ( [ eq - probcurr ] ) is useful for calculations reported in the sections below . the specific initial condition and the boundary conditions that we impose are motivated by the physical situation that our models are intended to capture . if initial value of @xmath5 is @xmath32 , the corresponding initial condition for ( [ main ] ) can be expressed as @xmath33 throughout this paper we choose @xmath34 , i.e. , maximum overlap between the outer surface of the mt and innser surface of the coaxial cylindrical sleeve - like coupler . the lifetime of an arbitrary kt - mt attachment is defined here as the time @xmath35 taken by the hypothetical particle , that is initially at @xmath4 , to reach @xmath3 _ for the first time_. thus , @xmath35 is essentially a first passage time @xcite that fluctuates from one attachment to another . the average value @xmath36 is the mean life time of the attachment . although @xmath34 initially , @xmath5 does not necessarily decrease monotonically . at any arbitrary instant of time @xmath5 can also increase , just as it can decrease , following the dynamical equations provided @xmath37 . one advantage of the unique initial condition @xmath34 is that the random variation of the lifetime of the kt - mt attachment from one run to another arises strictly from the intrinsically random kinetics and not from the choice of any random initial condition . the distribution of the lifetimes of kt - mt coupler that we have derived does not require any further averaging . in contrast , if a random starting location were selected , one extra averaging ( over sufficiently large number of random initial locations ) would be required to get a meaningful distribution of the lifetimes . since the mt is not allowed to penetrate the kinetochore plate , and the length of the sleeve - like coupler is @xmath1 , the overlap @xmath5 can not be large than @xmath1 . therefore , at @xmath4 we impose a reflecting boundary condition @xmath38 which implies @xmath39 an absorbing boundary condition @xmath40 is imposed so that spontaneous re - formation of the kt - mt is not possible and the time taken for @xmath5 to attain this boundary , starting from its initial value @xmath41 is the first - passage time that we define as the life time of the attachment . ( a ) , for a few different values of @xmath18 , ( in ( a ) ) , and @xmath29 ( in ( b ) ) . the parameter values used for this plot are @xmath42s@xmath43 , @xmath44s@xmath43 . ] ( b ) , for a few different values of @xmath18 , ( in ( a ) ) , and @xmath29 ( in ( b ) ) . the parameter values used for this plot are @xmath42s@xmath43 , @xmath44s@xmath43 . ] we use the laplace inversion method @xcite for extracting the relevant quantitative information from equation ( [ main ] ) . defining the laplace transform of probability density by , @xmath45 equation ( [ main ] ) can be re - expressed in terms of @xmath46 ; solving that equation under the given boundary conditions ( see appendix a for the details ) we get the exact analytical expression for @xmath46 . let @xmath47 denote the probability density , in the laplace space @xmath48 , for the first passage times to reach the position @xmath5 , given that the initial position was @xmath32 . using the relation @xcite @xmath49 between @xmath46 and @xmath47 , and the initial condition @xmath34 , we get @xmath50 } \label{main2}\ ] ] where @xmath51 includes contributions from both the drift velocity @xmath52 and the velocity - like quantity @xmath53 that arises from diffusion . for the calculation of the mean we do not need the distribution in @xmath0-space . instead these can be derived from the relation @xmath54 using the expression ( [ main2 ] ) for @xmath55 , we get @xmath56 an interesting feature of eqn.([eq - meant ] ) is that @xmath57 is like an effective energy barrier ; kt - mt detachment is achieved by crossing this barrier . since in addition to the energetic contributions from @xmath29 and @xmath8 , the barrier @xmath57 gets contribution also from @xmath13 and @xmath14 , this effective barrier is , at least partly , of kinetic origin . external tension @xmath8 influences the kt - mt attachment time @xmath58 in two possible ways . + ( 1 ) @xmath8 decreases the depth of the linear potential well which effectively reduces the mean attachment time . ( 2 ) the same @xmath8 also reduces the depolymerization rate @xmath15 thereby increasing the mean attachment time . + these two effects of the same load tension @xmath8 act against each other . the second effect can dominate over the first at small @xmath8 , thereby giving rise to nonmonotonic variation of @xmath58 with @xmath8 , provided @xmath15 falls sharply with increasing @xmath8 ; such a possibility is ensured if @xmath18 is sufficiently small . therefore , as shown graphically in fig.[mean](a ) , @xmath58 varies nonmonotonically with @xmath8 for both @xmath59pn and @xmath60pn . but , for @xmath61pn , the fall of @xmath15 is not sharp enough to compensate the reduction of the mean attachment time caused directly by @xmath8 . thus , the physical origin of the nonmonotonic variation of @xmath58 with @xmath8 is explained by fig.[mean](a ) in a transparent manner . the nonmonotonic variation of @xmath58 with @xmath8 is displayed also for all the three different values of the parameter @xmath29 in fig.[mean](b ) ; for any given @xmath8 , the larger is the @xmath29 ( i.e. , the deeper is the potential well ) , the longer is the mean life time of the kt - mt attachment . using the symbol @xmath62 to denote the tension that corresponds to the maximum of the average lifetime , from ( [ eq - meant ] ) we get @xmath63 the dependence of @xmath62 on @xmath18 , obtained from ( [ eq - fmax ] ) , is shown in fig.[fig - fmax ] . interestingly , the variation of @xmath62 with @xmath18 is non - monotonic . in the limit @xmath64 , @xmath65 for all @xmath66 . on the other hand , as @xmath67 , @xmath68 for all finite @xmath8 . therefore , in both these limits , @xmath8-dependence of @xmath14 drops out and kt - mt attachment behaves like a slip - bond . the monotonic decrease of @xmath62 with @xmath18 in fig.[mean ] arises from the fact that we have not plotted @xmath58 for even smaller values of @xmath18 where @xmath18 increases with @xmath18 because the corresponding values of @xmath58 turned out to be unphysically long . , the tension that corresponds to the maximum if @xmath58 , is plotted against @xmath18 . ] at this stage we can scrutinize the necessity of the exponential form ( [ eq - fdepbeta ] ) for the tension - dependence of the mt depolymerization rate @xmath14 . as shown in fig.[mean](a ) , a catch - bond - like behavior follows if @xmath15 decreases sufficiently sharply with increasing @xmath8 . on the other hand , if @xmath15 falls very slowly with increasing @xmath8 the system exhibits a slip - bond - like behavior . the exponential decrease assumed in equation ( [ eq - fdepbeta ] ) is an ideal choice for the @xmath8-dependence of @xmath15 because the crossover from catch - bond to slip - bond can be displayed by simply varying @xmath18 which determines the sharpness of the decrease @xmath15 . the series representation of equation ( [ eq - meant ] ) @xmath69 \label{eq - approxmeant}\ ] ] is exact . in the special case @xmath70 eqn.([eq - approxmeant ] ) reduces to @xmath71 which , as expected , arises from kt - mt detachment caused by pure diffusion . the corrections to the diffusive result in the limit @xmath72 , can be estimated from eqn.([eq - approxmeant ] ) . as we clearly stated in the introduction , the aim of the minimal model ( and , to some extent , even the more detailed model ) is not quantitative comparison with the experimental data reported by akiyoshi et al . instead , our aim is to present analytical calculations for a minimal model that provides physical insight into the interplay of opposing forces and competing kinetic processes that , together , determine the stability of the kt - mt coupler . nevertheless , for graphical plots , which often provide a more intuitive understanding , one needs numerical values of the parameters . the largest strength of the load tension applied by akiyoshi et al . @xcite was about @xmath73 pn while the lifetime peaked closed to @xmath74pn . the longest life time that they measured was about @xmath75 minutes . although the mean lifetimes plotted in fig.[mean](a ) are comparable to the measured values for at least one set of parameter values ( corresponding to @xmath76pn ) , the numerical values of the corresponding load tension @xmath8 are much smaller than those applied in the experiment of akiyoshi et al . however , this discrepancy does not invalidate our theory . the magnitudes of the forces in our model depend on the numerical values of the parameters like , for example , @xmath29 , @xmath77 , etc . on the other hand , the corresponding effective values in the experiments of akiyoshi et al.@xcite are not known . therefore , in our plots we have used parameter values that yield lifetimes comparable to those reported in experiments . most probably , our minimal model does not capture all the details of the in - vitro study @xcite . even so , we note that the inclusion of our two postulates is sufficient to generate non - monotonic variation of the lifetimes with load tension . we begin by pointing out that in the continuum formulation treated above , the expression ( [ eq - probcurr ] ) for the probability current density @xmath78 can be recast as @xmath79 with @xmath80 where the effective potential @xmath81 is given by @xmath82 x.\ ] ] note that @xmath83 accounts for the drift caused by the force @xmath84 as well as that arising from the polymerization - depolymerization kinetics of the mt . in this section , we utilize the effective potential @xmath83 to simulate the stochastic movements of kinetochore by using a method of discretization popularized by wang , peskin and elston ( from now onwards referred to as wpe method ) @xcite . following wpe prescription , we discretized the space into @xmath85 cells , each of length @xmath86 ; however , @xmath87 need not be identical to @xmath88 , the separation between the two consecutive binding site of a mt ( see fig.[fig - discretize ] ) . accordingly , the continuous effective potential @xmath83 is replaced by its discrete counterpart @xmath89 x_{j } \label{eq - discreteu}\ ] ] where @xmath90 denotes the position of the center of the @xmath91-th cell . is discretized into @xmath85 cells each of length @xmath92 following the wpe prescription ( see text for details ) . ] next we approximate the continuous movement described by the fokker - planck equation ( [ main ] ) by a master equation in terms of discrete jumps from the center of one cell to that of an adjacent cell either in the forward or backward direction . in this case the expressions for the forward and backward transition rates @xmath93 and @xmath94 ( see fig.[fig - discretize ] ) are given by @xcite @xmath95 @xmath96 where @xmath97 for the simple potential ( [ eq - discreteu ] ) , @xmath93 and @xmath94 are independent of site index @xmath91 . + in computer simulations , the mt is fully inserted into the sleeve in the initial state . the simulation produce the numerical data for the first passage time of the mt tip to exit the coupler . in fig . [ hist ] , histograms of these numerical data are plotted for a few values of force @xmath8 . laplace transform of these distributions are compared in the insets with the corresponding prediction of the exact analytical formula ( [ main2 ] ) . excellent agreement between the theoretical prediction and the simulation data ( albeit in the laplace space ) demonstrates how the theory may be useful in analyzing also the distribution of experimentally measured lifetimes of kt - mt attachments _ in - vitro_. in this section we begin with the assumption that the inner surface of the cylindrically shaped coupler consists of only passive binders . we assume that each binder head engages with the mt by obeying a unit energy function @xmath98 ( see fig.[fig2a ] ) , which has two key parameters : @xmath99 measures free energy drop due to binder affinity for the mt lattice , and @xmath100 describes the activation barrier for transitions between specific mt lattice binding sites @xcite . in other words , @xmath99 is a measure of the strength of the mt - binder affinity while @xmath100 is a measure of the `` roughness '' or friction of the mt - coupler interface . the total potential energy function is given by ( see fig.[fig2a ] ) @xmath101 where @xmath48 is the spacing between consecutive coupler binders ( see fig [ fig2a ] ) . binder spacing is an arbitrary parameter . here we set @xmath102 , where @xmath88 is the distance between consecutive mt binding sites . and @xmath103 ) . the load tension @xmath8 tilts the potential energy landscape @xmath103 . the tension - dependent depolymerization rate @xmath15 increases the effective barrier along the unbinding pathway ( orange line , inset ) by slowing down depolymerization . ] t , b = 0.001a . intermediate regime ( circle ) a = 0.5k@xmath104 t , b = 0.04a , and intermediate regime ( square ) a = 0.6k@xmath104 t , b = 0.2a . high - friction regime ( dashed red line ) is obtained from eq . ( 32 ) with @xmath105 , @xmath106 , @xmath107 pn . ] for @xmath103 in our numerical calculations , we have used the simple expression @xmath108 where @xmath109 , @xmath110 and @xmath111 . the linear and scalar coefficients in eq . ( [ eq5 ] ) arise because we use a fourier series to approximate the well function expression . thus , the mt - coupler interaction is represented more realistically by @xmath103 than by @xmath10 of the minimal model . the friction arising from the potential landscape @xmath103 gives rise to a jump - drift - diffusion process . as we ll point out below , this general version of our model reduces to the minimal version discussed above , and hence effectively to a drift - diffusion process , in a special limiting case . in infinitesimal time interval @xmath112 , the change @xmath113 of the overlap is given by @xmath114dt+ \ell dn_{r}(t)+\sigma dw(t ) \label{eq1}\ ] ] where , as before , the constant @xmath8 is the external load tension and @xmath24 is the effective drag coefficient . @xmath115 accounts for the thermal diffusion of the coupler on the lattice with @xmath116 , and @xmath117 is a poisson counting process describing the mt tip dynamics with rate @xmath118 that is decided by the difference of the mt polymerization rate @xmath13 and mt depolymerization rate @xmath14 . the stochastic differential equation ( [ eq1 ] ) is essentially equivalent to the langevin equation ( [ eq - langevin ] ) ; the form ( [ eq1 ] ) is used in our numerical computations for the convenience of implementation . @xmath119 is characteristic of the structure of the coupler ( coupler length ) whereas its energetics depend on @xmath120 ( i.e. , on the parameters @xmath99 , @xmath100 ) and @xmath8 ; the stochastic kinetics are influenced by the interplay of forces arising from the potential landscape , random brownian forces , and by @xmath13 , @xmath14 . by a combination of standard analytical and numerical methods , we study the trends of variation of @xmath121 with ( a ) @xmath8 , as well as , the ( b ) size , ( c ) energetics , and ( d ) kinetics of the coupler . it is difficult to derive an exact analytical expression for the mean first passage time corresponding to eq . ( [ eq1 ] ) that would be valid in all parameter regimes . therefore , we explore two limiting cases for which explicit approximate solutions can be obtained : ( a ) _ low - friction regime _ ( @xmath122 ) , and ( b ) _ high - friction regime _ ( @xmath123 ) . in the _ low - friction regime _ the coupler can easily rearrange its position relative to the mt ; in this regime @xmath121 approximated well by the expression in equation ( [ eq - meant ] ) in the _ high - friction regime _ large @xmath100 leads to strong local pinning that practically stalls it . the only mode available for detachment of the coupler is via depolymerization of the mt . consequently , in this regime @xmath124 the exponential increase of @xmath121 with @xmath8 in the high - friction regime can be regarded as an artifact of the unphysical ( or , unrealizable ) nature of this regime . unless explicitly stated otherwise , the parameter values that we have used for our numerical calculations in this section are given in the table [ default ] . ccc parameter description & symbol & values tested + mt binding site spacing & @xmath88 & 8/13 nm @xcitemaximal coupler length & @xmath1 & 50 nm @xcitemaximal number of coupler binders & @xmath119 & 15 - 65 @xcitepolymerization rate & @xmath13 & @xmath125 @xmath126@xcitemaximal depolymerization rate & @xmath20 & @xmath127 @xmath126 @xcitecritical depolymerization force & @xmath18 & @xmath128 pn ( estimated here)external load force & @xmath8 & @xmath129 pn force of attractive kt - mt interaction & @xmath29 & @xmath130 pn mt lattice / binder binding energy & @xmath99 & @xmath131 @xcitebinder activation barrier & @xmath100 & @xmath132 ( estimated here)diffusion constant & @xmath22 & @xmath133 nm@xmath134s@xmath43 @xciteeffective drag coefficient & @xmath24 & @xmath135 pns@xmath136m@xmath43@xcite [ default ] . ( a)-(b ) @xmath121 for two sets of maximal rates of mt depolymerization with the ratio @xmath137 fixed . in ( a ) the data are obtained for an _ intermediate _ regime coupler , with @xmath138 k@xmath104 t , @xmath139 , and in ( b ) for a shorter coupler in the _ low - friction _ regime with @xmath140 k@xmath104 t , @xmath141 . ( c)-(d ) @xmath121 for _ intermediate _ regime couplers . in ( c ) are shown data with two different strengths of friction caused by two values of @xmath100 , and in ( d ) for two different binding energies given by @xmath99 . the mt rates are @xmath142 s@xmath43 , @xmath143 s@xmath43 for ( c)-(d ) . ( e ) @xmath121 plotted against @xmath8 for three different lengths of the _ low - friction _ coupler from ( a ) with @xmath140 k@xmath104 t , @xmath144 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . error bars mark standard deviation.,title="fig : " ] + . ( a)-(b ) @xmath121 for two sets of maximal rates of mt depolymerization with the ratio @xmath137 fixed . in ( a ) the data are obtained for an _ intermediate _ regime coupler , with @xmath138 k@xmath104 t , @xmath139 , and in ( b ) for a shorter coupler in the _ low - friction _ regime with @xmath140 k@xmath104 t , @xmath141 . ( c)-(d ) @xmath121 for _ intermediate _ regime couplers . in ( c ) are shown data with two different strengths of friction caused by two values of @xmath100 , and in ( d ) for two different binding energies given by @xmath99 . the mt rates are @xmath142 s@xmath43 , @xmath143 s@xmath43 for ( c)-(d ) . ( e ) @xmath121 plotted against @xmath8 for three different lengths of the _ low - friction _ coupler from ( a ) with @xmath140 k@xmath104 t , @xmath144 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . error bars mark standard deviation.,title="fig : " ] + . ( a)-(b ) @xmath121 for two sets of maximal rates of mt depolymerization with the ratio @xmath137 fixed . in ( a ) the data are obtained for an _ intermediate _ regime coupler , with @xmath138 k@xmath104 t , @xmath139 , and in ( b ) for a shorter coupler in the _ low - friction _ regime with @xmath140 k@xmath104 t , @xmath141 . ( c)-(d ) @xmath121 for _ intermediate _ regime couplers . in ( c ) are shown data with two different strengths of friction caused by two values of @xmath100 , and in ( d ) for two different binding energies given by @xmath99 . the mt rates are @xmath142 s@xmath43 , @xmath143 s@xmath43 for ( c)-(d ) . ( e ) @xmath121 plotted against @xmath8 for three different lengths of the _ low - friction _ coupler from ( a ) with @xmath140 k@xmath104 t , @xmath144 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . error bars mark standard deviation.,title="fig : " ] + , is plotted in the low - friction regime and the corresponding data for the intermediate regime . the parameters are @xmath119 = 65 , a = 0.45 k@xmath104 t , @xmath145 s@xmath43 , @xmath146 s@xmath43 . error bars mark standard deviation.,width=350,height=254 ] for numerical simulations of this version of our model in the intermediate regime we used the parameter values which are within the range reported in the literature@xcite . consequently , computed @xmath36 turn out to be comparable to those observed in the experiments . some of the typical curves are plotted in figs.[fig3 ] . @xmath147 _ effects of variation of rate of kinetics _ the effects of varying the maximal rate of depolymerization @xmath20 , keeping the ratio @xmath148 fixed , is demonstrated in figs.[fig3](a ) and ( b ) . clearly , in such situations , the kt - mt attachment can exhibit catch - bond - like or slip - bond - like behavior depending on the value of @xmath20 . @xmath147 _ effects of variation of parameters that characterize energetics _ the effects of the energetics , namely the influences of the parameters @xmath99 and @xmath100 that determine the potential energy landscape , are displayed in figs.[fig3](c ) and ( d ) . the effects of varying the number of binders @xmath119 is shown in fig.[fig3](e ) . in all these cases the nonmonotonic variation of @xmath121 with @xmath8 seems to be a generic feature , except that @xmath121 may become too large to be observable for some values of the parameters ( see , for example , fig.[fig3](e ) . @xmath147 _ tension - induced switching from depolymerization to polymerization _ fig.[fig3d ] shows the key role of the external tension @xmath8 in deciding the direction of the velocity . suppose the parameter values are chosen such that @xmath20 dominates over @xmath13 in the absence of external tension thereby ensuring depolymerization of the mt . when the external tension is applied , the depolymerization is suppressed . increasing @xmath8 can eventually cause such strong suppression of the depolymerization that , beyond a certain strength of @xmath8 , @xmath13 dominates over @xmath15 . in such a scenario , the mt switches from the depolymerization mode to polymerization mode at a threshold strength of @xmath8 . in this paper , we have presented theoretical models of kt - mt attachment at three levels of detail starting with a minimal version that captures only the key components and essential processes . these models mimic kt - mt attachments reconstituted _ in - vitro _ from purified components extracted from budding yeast @xcite . exploiting the simplicity of the minimal version of the model , we have calculated the distribution of the lifetimes of the kt - mt attachments analytically and compared the results with the corresponding simulation data . equation ( [ main2 ] ) encapsulates the main result of this part of the manuscript . the strength of the minimal model is that the results can be derived analytically . in spite of its simplicity , this minimal model provides deep insight into the roles of opposing forces and competing kinetics in the tension - dependence of the life time of the kt - mt attachments . the second version of the model is an extension of the minimal model ; it incorporates some more realistic features of the interaction between the mt and the coupler . for the second model we have succeeded in calculating analytically only the mean lifetime of the kt - mt attachments under reasonable schemes of approximation . the analytical expression ( [ eq - meant ] ) for the mean life time of the kt - mt attachment in the minimal model has been derived from the corresponding full distribution ( [ main2 ] ) of the life times , the nonmonotonic variation of the mean lifetime of the kt - mt attachments with external tension exhibited over wide range of values of the model parameters in both the models is consistent with the similar trend observed in the _ in - vitro _ experiments @xcite . the realistic potential landscape in the second model does not alter this qualitative trend . we have also demonstrated that the nonmonotonicity of the variation of @xmath58 with @xmath8 depends on the rapidity of decrease of @xmath15 with increasing @xmath8 . by varying @xmath18 , which determines the sharpness of the decrease of @xmath15 , we have shown how the catch - bond - like behavior crosses over to a slip - bond - like behavior . moreover , our analytical calculations also show that , unlike other conventional catch - bonds , the catch - bond - like behavior of the kt - mt attachment arises from the interplay of not only forces derivable from the potential landscape , but also by a competition of the kinetics of polymerization and depolymerization of the mt . in fact , the barrier against the breakdown of the attachment gets contributions from both opposing forces and competing kinetic processes . the slip - bond - like monotonic decrease of @xmath121 with @xmath8 observed in some of the parameter regimes ( whose physical origin has been explained in terms of the nature of @xmath8-dependence of the depolymerization rate @xmath15 ) , might be detectable in experiments under conditions different from those used by akiyoshi et al.@xcite . as expected , we have found that the molecular motors enhance stability or tend to destabilize the kt - mt attachment depending on whether the force it generates opposes or assists the external tension . the latter awaits experimental confirmation . moreover , our analytical predictions on the distribution of the lifetimes will be useful in the future in analyzing the statistical properties of the lifetimes provided the _ in - vitro _ experiment on the reconstituted kt - mt attachment is repeated sufficiently large number of times . in principle , the models presented here can be further extended by including active force generators like molecular motors @xcite . however , so far such active force generators have not been detected at the kt - mt interface in budding yeast although such force - generating motors have been found in other cells including , for example , mammalian cells . but , in contrast to budding yeast , where each kinetochore can attach to only one mt , multiple mts can attach with each kinetochore in a mammalian cell . since modeling of kt - mt attachment in this paper is focussed almost exclusively on budding yeast , we have not discussed the effects of molecular motors in the main text . thus , the extended model in appendix b is not a complete model of kt - mt attachment in mammalian cells ; it is intended to indicate how the effects of molecular motors can be incorporated within the broad theoretical framework developed in this paper . in near future we hope to extend our model to kt - mt attachments in mammalian cells . however , at present , the main hurdle in this modeling is the lack of our current understanding as to how multiple mts attached to the same kt coordinate their kinetics @xcite . the `` sleeve model '' developed originally by hill @xcite was adapted by joglekar and hunt @xcite , to account for the phenomenon of `` directional instability '' @xcite ( also referred to as chromosome oscillation ) observed during mitosis in vertebrates . in those cells each kt can attach with several mt ( typically , up to 35 ) and this feature was also incorporated in the model studied by joglekar and hunt by carrying out computer simulations . our work differs significantly from that of joglekar and hunt @xcite . first , our model is intended to account for the qualititative features of the results obtained from a much simpler _ in - vitro _ system where only one mt is attached to only a single kt that is subjected to an externally applied load force that , in our model , is assumed to alter the rate of mt depolymerization itself . moreover , because of their motivation in studying directional instability , joglekar and hunt @xcite put major emphasis on monitoring the _ spatial _ displacements that characterize this instability . in contrast , the quantity of our main interest is the life _ time _ of the kt - mt attachments . the spatial excursions of the kinetochores was also the main quantity of interest of gardner et al.@xcite who studied the effects of tension - mediated regulation of the kinetics of kinetochore mts by computer simulation of a related model . the scope of the more detailed model studied by scholey et al.@xcite was even broader . our force - balance equations are similar , at least in principle , to the force - balance equations formulated by scholey et al.@xcite . however , instead of the spatial displacements of the kinetochores , the lifetimes of the kt - mt attachments are the main focus of our investigation . most of the earlier models were based on the assumption that the coupler is a stable sleeve or ring . in recent years , a new class of models have been developed on the basis of an altogether different assumption regarding the nature of the coupler . in this scenario @xcite microtubule associated proteins ( maps ) are assumed to make transient attachments with the mt ( i.e. , each map attaches to the mt for a brief duration before getting detached from it ) . the entropic contribution to the force is significant and slight variation in the mean lifetime of the individual transient attachments of the maps can have large effect on the lifetime of the kt - mt attachment . in order to account for their experimental observations , akiyoshi et al . @xcite developed a 2-state kinetic model . this model was motivated by the formal analogy with the models of ` catch - bonds ' in other systems . the growing and shrinking states of the mt were argued to be analogs of the strongly- and weakly - bound states in the models of catch - bonds . in their 4-parameter kinetic model , two parameters were the rates of transitions between the states of growth and shrinkage of the mt . the remaining two parameters were the rates of detachment of the kinetochore from the mt while the latter is in the growing and shrinking phases , respectively . postulating exponential dependence of the type @xmath149 for each of the four rate constants ( n=1,2,3,4 ) , on the external tension @xmath8 , akiyoshi et al . extracted the numerical values of the 8 parameters @xmath150 ( n=1,2,3,4 ) corresponding to the best fit to their experimental data @xcite . in their model , akiyoshi et al.@xcite did not explicitly treat the kinetics of the polymerization and depolymerization , by attachment and detachment of the successive subunits ( more precisely , the @xmath151 tubulin dimers ) , during the growing and shrinking phases of the mt , respectively . therefore , the force - dependence of the four rate constants that provide the best fit to their experimental data do not indicate the corresponding force - dependences of the rates of polymerization and depolymerization of the mt . moreover , the potential energy of interaction between the mt and the coupler as well as the effect of the external tension on this potential energy landscape are not incorporated explicitly in their model . thus , the model of akiyoshi et al.@xcite does not directly demonstrate the interplay of the opposing forces and competing kinetics . since neither any structural features of the coupler , nor the nature of its interaction with the mt enters explicitly into akiyoshi et al.s @xcite model , their model can not be used to study the effects of the size and composition of the coupler or that of the nature of the mt - coupler interaction . neither can it be used to account for the special features of hybrid couplers . in contrast , in this paper we have used our models to study the effects of varying ( i ) the size of the coupler , ( ii ) relative population of active force generators ( molecular motors ) and passive binders , as well as ( iii ) the depth and roughness of the potential of interaction between the coupler and the mt . catch bond formed by the fimbriae of e - coli bacteria with target cell surface and that formed by eukaryotic p - selectin and integrin receptors with their ligands have been studied extensively in recent years @xcite to understand their mechanism . in this paper we have focussed exclusively on the kt - mt attachments to understand the mechanism of its catch - bond - like behavior @xcite using simple theoretical models . one distinct feature of the mt is that , unlike other ligands , these stiff filaments are capable of generating push and pull by their polymerization and depolymerization , respectively . in particular , the pulling force generated by a mt is powered by the hydrolysis of gtp bound to the tubulin subunits . as we have shown in this paper , the competing polymerization and depolymerization of mt gives rise to a unique feature of the catch - bond - like behavior of kt - mt attachments where the stable state can be a non - equilibrium ( local ) steady state rather than a local minimum of the free energy of the ligand - receptor system . we hope the modeling methodology developed here can be usefully adapted to understand the catch - bond - like behavior of depolymerizing actin under load tension observed recently by carrying out _ in - vitro _ experiments with atomic force microscope ( afm ) @xcite . * acknowledgements : * dc thanks frank jlicher for valuable comments . this work has been supported in part by csir ( india ) through a senior research fellowship ( aks ) , by nsf through the grant dms- 1358932 ( bs ) , by mbi , the ohio state university , through the nsf grant dms 0931642 ( bs and dc ) , by iit kanpur through dr . jag mohan garg chair professorship ( dc ) , by serb ( india ) through j.c . bose national fellowship ( dc ) , by dbt ( india ) through a research grant ( dc ) , and by the visitors program of the max - planck institute for the physics of complex systems , dresden ( dc ) . from equation ( [ main ] ) we get @xmath152 for @xmath153 , equation ( [ main1 ] ) has following solution . @xmath154 similarly , for @xmath155 , solution of equation ( [ main1 ] ) is , @xmath156 where @xmath157 and @xmath158 are given by @xmath159 with @xmath160 . imposing the absorbing boundary condition at @xmath3 on @xmath46 we find @xmath161 now @xmath46 and it s first derivative must satisfy the following matching conditions at @xmath162 : + @xmath163 @xmath164 reflecting boundary condition at @xmath4 gives , @xmath165 the four unknowns @xmath166 , @xmath167 , @xmath168 and @xmath169 can be obtained by solving the three equations ( [ b1])-([eq - alg3 ] ) . as we ll see later , we need @xmath46 only for @xmath170 . therefore , we give the expression for @xmath166 only in the region @xmath153 : @xmath171 with @xmath172 . and @xmath167 can be find by using equation [ b1 ] . thus , for @xmath153 @xmath173\ ] ] where the @xmath48-dependence enters through the @xmath48-dependence of @xmath174 ( see eqn.([eq - defnu ] ) that enters into the expression ( [ eq - defnv ] ) for @xmath175 . so far no motor protein has been detected at the kinetochore in budding yeast , which is the object of our modeling in this paper , but , in most of the eukaryotes ( e.g. , mammalian cells ) such motors have been found in the kinetochores each of which , however , attaches to more than one mt . on the other hand , not more than a single mt can attach to each kt in budding yeat . nevertheless , we extend our model further by including motor proteins @xcite , which are active force generators , in addition to the passive binders in the model coupler . we hope to extend this hybrid coupler , in near future , by integrating more than one mt to model kt - mt attachments of mammalian cells . as indicated in recent experiments @xcite , the outermost layer of the _ hybrid _ coupler in mammalian cells consists of passive binders while the innermost layer , that is adjacent to the kt , is composed of active force generators . in the model of hybrid coupler we represent the active force generators by molecular motors ( see fig [ fig4]a ) . as we show in this section , the nature of the response of the kt - mt attachment to the external tension now depends on whether the motors are plus - end directed ( e.g. , kinesin ) or minus - end directed ( e.g. , dynein ) . we use the subscripts or superscripts @xmath176 and @xmath177 to refer to the plus - end and minus - end directed motors , respectively . following the general trend in the literature on molecular motors @xcite . we also postulate a linear force - velocity relation for the individual motors : @xmath178 where @xmath179 and @xmath180 are the stall force and maximal velocity for the plus - end directed and minus - end directed motors , respectively , whereas @xmath181 are the corresponding instantaneous velocities . for an _ intermediate _ regime hybrid coupler with ( @xmath182 ) and without ( @xmath183 ) minus - end motors with @xmath184 k@xmath104 t , @xmath185 , @xmath142 s@xmath43 , @xmath143 s@xmath43 . ( c ) breaking load , f@xmath186 for a low - friction hybrid coupler versus motor densities ; minus - end motors ( squares ) and plus - end motors ( circles ) . average speed vs force for varying densities of minus - end motors . stall forces for each motor density fraction . parameters for ( c)-(d ) are @xmath187 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . for all panels @xmath188.,title="fig:",width=331,height=230 ] for an _ intermediate _ regime hybrid coupler with ( @xmath182 ) and without ( @xmath183 ) minus - end motors with @xmath184 k@xmath104 t , @xmath185 , @xmath142 s@xmath43 , @xmath143 s@xmath43 . ( c ) breaking load , f@xmath186 for a low - friction hybrid coupler versus motor densities ; minus - end motors ( squares ) and plus - end motors ( circles ) . average speed vs force for varying densities of minus - end motors . stall forces for each motor density fraction . parameters for ( c)-(d ) are @xmath187 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . for all panels @xmath188.,title="fig : " ] for an _ intermediate _ regime hybrid coupler with ( @xmath182 ) and without ( @xmath183 ) minus - end motors with @xmath184 k@xmath104 t , @xmath185 , @xmath142 s@xmath43 , @xmath143 s@xmath43 . ( c ) breaking load , f@xmath186 for a low - friction hybrid coupler versus motor densities ; minus - end motors ( squares ) and plus - end motors ( circles ) . average speed vs force for varying densities of minus - end motors . stall forces for each motor density fraction . parameters for ( c)-(d ) are @xmath187 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . for all panels @xmath188.,title="fig : " ] for an _ intermediate _ regime hybrid coupler with ( @xmath182 ) and without ( @xmath183 ) minus - end motors with @xmath184 k@xmath104 t , @xmath185 , @xmath142 s@xmath43 , @xmath143 s@xmath43 . ( c ) breaking load , f@xmath186 for a low - friction hybrid coupler versus motor densities ; minus - end motors ( squares ) and plus - end motors ( circles ) . average speed vs force for varying densities of minus - end motors . stall forces for each motor density fraction . parameters for ( c)-(d ) are @xmath187 , @xmath145 s@xmath43 , @xmath146 s@xmath43 . for all panels @xmath188.,title="fig : " ] we start with the force balance equation which does not include random fluctuations for the overlap velocity @xmath189 where @xmath24 is the effective coupler drag coefficient and @xmath190 is the velocity of the mt tip with respect to a space - fixed frame of reference . the active force term @xmath191 with the motor density function @xmath192 where @xmath193 is the standard heaviside step function and @xmath194 nm , corresponds to the total horizontal length of the coupler that can be populated by active components ( in three - dimensions this corresponds to one layer of motors working around a mt with 13 protofilament tracks , with one motor per track ) . note that in the symbol @xmath195 , the subscript minus ( - ) / plus ( + ) denote the force generated by the minus - end / plus - end directed motors that tend to increase / decrease the overlap @xmath5 . we have expression ( [ eq - fplusminus ] ) for @xmath195 where @xmath179 and @xmath180 are the stall force and maximal velocity for the plus - end directed and minus - end directed motors , respectively , whereas @xmath181 are the corresponding instantaneous velocities . next we express @xmath181 in terms of @xmath196 . using @xmath197 , in case of minus - end directed motors @xmath198 similarly , if only plus - end - directed motors are present , @xmath199 the coupler movement described by eq . ( [ deteqn ] ) is fully deterministic . next , we write down a stochastic differential equation ( sde ) that would , upon averaging , correspond to the deterministic equations written above . suppose over a small time interval @xmath204 the number of subunits ( an @xmath151 tubulin dimer ) added and removed from the tip of the mt by polymerization and depolymerization are @xmath205 , an independent homogenous poisson process . we capture the effects of random brownian forces through the noise @xmath206 which is assumed to be a gaussian stochastic process , which also includes the effects of fluctuations both in the chemical reactions and mechanical stepping involved in each cycle of the individual motors . thus , finally , the equation for the hybrid coupler overlap reads @xmath207dt\nonumber\\ & + \ell dn_{r}(t)+\sigma dw(t).\end{aligned}\ ] ] in fig . [ fig4 ] we plot results for a hybrid coupler with varying density fractions of motors at the active interface , measured by @xmath208 . the non - monotonic nature of @xmath121 is preserved when the active components are added ( see fig . [ if the active force generators are minus - end directed ( for example , dyneins ) , the force exerted by them oppose load force . consequently , the mean lifetime @xmath121 os the kt - mt attachment increases as motor fractions , @xmath209 become larger ( see figs . [ fig4]c ) . for a given @xmath209 , the extent of such motor - induced extra stabilization of hybrid couplers can be sensitive to the binding energy of the passive components measured by @xmath99 ( see fig [ fig4]c ) . despite enhancing the stability , higher numbers of load opposing motors can also lower the ability of the coupler to efficiently track rescued mt tips , as noted by the slower coupler velocities in fig . [ fig4]d . plus - end directed motors ( for example , cenp - e kinesin ) , which oppose overlap between the mt and the coupler , generate force that assist the external load . if the active force generators in the hybrid coupler are such plus - end directed motors , neither the coupler velocities and nor the breaking force @xmath210 show significant variation as @xmath209 is varied under a given tension , provided that there are sufficient numbers of passive binder components supporting attachment . 00 h. lodish , a. berk , s l. zipursky , p. matsudaira , d. baltimore , and j. darnell , _ molecular cell biology , 4th edition _ , ( w. h. freeman , 2000 ) . i. m. cheeseman and a. desai , nature rev . * 9 * , 33 ( 2008 ) . e. karsenti and i. vernos , science * 294 * , 543 ( 2001 ) . mcintosh , m.i . molodtsov and f.i . ataullakhanov , d. chowdhury , phys . rep . * 529 * , 1 ( 2013 ) . d. chowdhury , biophys . j. * 104 * , 2331 ( 2013 ) . g. civelekoglu - scholey , d.j . sharp , a. mogilner and j.m . scholey , biophys . j. * 90 * , 3966 ( 2006 ) . joglekar and a.j . hunt , biophys . j. * 83 * , 42 ( 2002 ) . gardner , c.g . pearson , b.l . sprague , t.r . zarzar , k. bloom , e.d . salmon and d.j . odde , mol . cell * 16 * , 3764 ( 2005 ) . quart . * 45 * , 147 - 207 ( 2012 ) . c.s . peskin and g.f . oster , biophys . j. * 69 * , 2268 ( 1995 ) . b. akiyoshi , k. k. sarangapani , a. f. powers , c. r. nelson , s. l. reichow , h. arellano - santoyo , t. gonen , j. a. ranish , c. l. asbury and s. biggins , nature * 468 * , 576 ( 2010 ) . , ann rev biophys , * 37 * , 399 - 416 ( 2008 ) . h. wang , c. peskin and t. elston , j. theo . biol . * 221 * , 491 ( 2003 ) . h. wang , t. c. elston , j. stat . phys . * 128 * , 35 ( 2007 ) . t. hill , proc . * 82 * , 4404 ( 1985 ) . s. westermann , d.g . drubin and g. barnes , annu . rev . biochem . * 76 * , 563 ( 2007 ) . asbury , j.f . tien and t.n . davis , trends in cell biol . * 21 * , 38 ( 2011 ) . buttrick and j.b.a . millar , chromosome res . * 19 * , 393 ( 2011 ) . e.a . foley and t.m . kapoor , nat . . cell biol . * 14 * , 25 ( 2013 ) . a. d. franck , a. f. powers , d. r. gestaut , t. gonen , t. n. davis , c. l. asbury , nat cell biol * 9 * , 832 ( 2007 ) . s. redner , _ a guide to first - passage processes _ , ( cambridge university press , 2001 ) . v. balakrishnan , _ elements of nonequilibrium statistical mechanics _ , ( ane books pvt . ltd . , 2008 ) b. shtylla and j. p. keener , siam j. appl . math * 71 * 1821 ( 2011 ) . a. p. joglekar , d. c. bouck , j. n. molk , k. s. bloom and e. d. salmon , nat . cell biol . * 8 * , 581 ( 2006 ) . a. p. joglekar , k. bloom and e. d. salmon , curr . biol . * 19 * , 694 ( 2009 ) . x. wan , r. p. oquinn , h. l. pierce , a. p. joglekar , w. e. gall , j. g. deluca , c. w. carroll , s.t . liu , t. j. yen , b. f. mcewen , p. t. stukenberg , a. desai and e.d . salmon , cell * 137 * , 672 ( 2009 ) . j. lawrimore , k. s. bloom , e. d. salmon , , j cell biol . * 195 * , 573 ( 2011 ) . sharp , g.c . rogers and j.m . scholey , nature * 407 * , 41 - 47 ( 2000 ) . j.m . scholey and a. mogilner , _ mitotic spindle motors _ , in : _ molecular motors _ , ed . m. schliwa ( wiley - vch , 2003 ) . h. maiato , j. deluca , e. d. salmon , and w. c. earnshaw , j. cell science * 117 * , 5461 ( 2004 ) . s. dumont , e. d. salmon , t. j. mitchison , science * 337 * , 355 ( 2012 ) . s. gonen , b. akiyoshi , m. g. iadanza , d. shi , n. duggan , s. biggins , t. gonen , nat struct mol biol * 2012 * , 925 ( 2012 ) . w. thomas , annu . * 10 * , 39 ( 2008 ) . thomas , curr . * 19 * , 50 ( 2009 ) . sokurenko , v. vogel , w.e . thomas , cell host & microbe * 4 * , 314 ( 2008 ) . o.v . prezhdo and y.v . pereverzev , acc . 42 * , 693 ( 2009 ) . evans and d.a . calderwood , science * 316 * , 1148 ( 2007 ) . c. zhu , j. lou and r.p . mcever , biorheology * 42 * , 443 ( 2005 ) . c. zhu , annals biomed 36 * , 604 ( 2008 ) . mcever and c. zhu , annu . * 26 * , 363 ( 2010 ) . p. sundd , m.k . pospieszalska , l.s.l . cheung , k. konstantopoulos and k. ley , biorheology * 48 * , 1 ( 2011 ) . p. sundd , m.k . pospieszalska and k. ley , mol . immunol . * 55 * , 59 ( 2013 ) .	before a cell divides into two daughter cells , chromosomes are replicated resulting in two sister chromosomes embracing each other . each sister chromosome is bound to a separate proteinous structure , called kinetochore ( kt ) , that captures the tip of a filamentous protein , called microtubule ( mt ) . two oppositely oriented mts pull the two kts attached to two sister chromosomes thereby pulling the two sisters away from each other . here we theoretically study an even simpler system , namely an isolated kt coupled to a single mt ; this system mimics an _ in - vitro _ experiment where a single kt - mt attachment is reconstituted using purified extracts from budding yeast . our models not only account for the experimentally observed `` catch - bond - like '' behavior of the kt - mt coupling , but also make new predictions on the probability distribution of the lifetimes of the attachments . in principle , our new predictions can be tested by analyzing the data collected in the _ in - vitro _ experiments provided the experiment is repeated sufficiently large number of times . our theory provides a deep insight into the effects of ( a ) size , ( b ) energetics , and ( c ) stochastic kinetics of the kt - mt coupling on the distribution of the lifetimes of these attachments .
we study theoretically the resistive state in short superconducting microbridges ( with length l less than the coherence length @xmath2 ) at temperatures close to the critical one . this subject was widely discussed and studied in the 70-ies of last century ( for review see @xcite ) and has regained renewed interest recently ( see for example refs . interesting phenomena occurring in such a system , and which are still not completely understood , are : i ) the hysteresis of current - voltage ( iv ) characteristics at relatively low temperatures @xcite and ii ) the foot-like ( sometimes also called shoulder-like - see ref . @xcite ) feature in the iv characteristics observed experimentally mainly in tin microbridges at low voltages and at temperatures close to @xmath3 @xcite ( see also @xcite ) . the hysteresis is usually explained by joule heating while for the foot-like structure several theories were proposed @xcite that are based on the idea that the quasiparticle distribution function is out of equilibrium ( overcooled ) as a consequence of the variation in time of the magnitude of the superconducting order parameter @xmath4 in the superconducting microbridge . the motivation for those theories comes from the fact that the energy of the quasiparticles depends on @xmath0 and when the characteristic time scale for the variation of @xmath0 is smaller than the inelastic relaxation time @xmath5 of the quasiparticle distribution function @xmath6 the occupancy of the states with energy @xmath7 may differ from the equilibrium one ( for a detailed discussion see the book of tinkham @xcite ) . to simplify the analytical treatment of the problem the authors of refs . @xcite assumed that in the dynamic ( resistive ) state @xmath0 varies as fast as @xmath8 and to find the coordinate and time dependence of @xmath9 they solved the _ stationary _ ginzburg - landau equation with a time dependent @xmath10 . furthermore they assumed that : 1)the relaxation term in the kinetic equations can be neglected when the period of oscillations of the order parameter is much smaller than @xmath5 , 2 ) the spectral functions depend only on the local magnitude of the order parameter @xmath0 ( so called local approach ) . we argue that for a realistic inelastic relaxation time @xmath5 , and even for a short microbridge @xmath11 the nonequilibrium contributions to @xmath6 strongly affects the dynamics of @xmath0 and results in a larger time scale for @xmath0 than for @xmath8 . moreover the relaxation term in the kinetic equations plays a very important role at any voltage and therefore can not be omitted . taken together with the dependence of the spectral functions on both @xmath0 and @xmath1 they provide an averaged heating of the quasiparticles , instead of cooling @xcite . some kind of cooling of quasiparticles at high voltages can be found only when one takes into account additional terms in the kinetic equations which couple the longitudinal @xmath12 ( odd in energy ) and transverse @xmath13 ( even in energy ) parts of @xmath14 due to the finite spectral supercurrent ( in previous works @xcite these terms were omitted ) . this cooling is not effective at low voltages and heating together with the large time scale of the variation of @xmath0 result in a hysteresis of the iv characteristics for relatively large @xmath5 . we should stress that the time averaged heating of quasiparticles is driven by oscillations of @xmath0 in the superconducting microbridge and _ not _ by joule dissipation @xmath15 ( * j * is the current density and * e * is the electric field ) . the paper is organized as follows . in section ii we discuss the theoretical model . in section iii we present and discuss our results . in section iv we present our conclusions and discuss the possible origin of the experimentally found @xcite foot-like structure in the iv characteristics . to simulate the resistive state in a short superconducting microbridge we use the kinetic equations derived in refs . @xcite for dirty superconductors near @xmath3 @xmath16 @xmath17 here @xmath18 is a quantity which is proportional to the superfluid velocity ( @xmath19 ) , @xmath20 is an electrostatic potential , @xmath21 and @xmath22 . @xmath23 , @xmath24 , @xmath25 are the spectral functions which should be found from the usadel equation for the normal @xmath26 and anomalous @xmath27 , @xmath28 ( @xmath29 ) green functions @xmath30 eqs . 1(a , b ) are coupled through the finite spectral supercurrent @xcite @xmath31 . below we show that the coupling terms in eqs . 1(a , b ) strongly influence @xmath32 and the dynamics of the order parameter which is described by the modified time - dependent ginzburg - landau equation @xmath33 where @xmath34 and @xmath35 are the zero temperature ginzburg - landau coherence length and the corresponding order parameter . nonequilibrium parts of the quasiparticle distribution function enters eq . ( 3 ) via the potentials @xmath36 and @xmath37 . when @xmath38 is negative the potential @xmath39 is positive and vice versus . in some respect , from the structure of eq . ( 3 ) it follows , that one may introduce an effective temperature for the quasiparticles @xmath40 and thus the positive / negative sign of @xmath41 means local heating/cooling of quasiparticles . we should stress that we use the term effective temperature only in order to give a simple physical interpretation of our numerical results and describe the _ integral _ effect of the nonequilibrium distribution @xmath6 which enters the equation for the order parameter via the potentials @xmath41 and @xmath42 . please note that the resulting nonequilibrium @xmath6 can not be viewed as a fermi - dirac function with effective temperature @xmath43 and electrostatic potential @xmath44 . the current and the electrostatic potential in the sample can be found using the following equations @xmath45 @xmath46 where @xmath47 is the normal state conductivity . in metals we have for the charge density @xmath48 and the condition @xmath49 is satisfied due to eqs . ( 1b , 3 , 5 ) . in the derivation of eqs . 1(a , b ) it was assumed that deviations from equilibrium are small @xmath50 . it allowed one to linearize the collision integral due to electron - phonon collisions and to write it in the relaxation time approximation . furthermore , it was assumed that the inelastic relaxation time due to electron - electron interactions is much larger than due to electron - phonon and hence one can neglect the corresponding collision integral . to model the superconducting microbridge connected to superconducting bulk leads in the current carrying regime we consider the system shown in fig . it consists of normal bulk reservoirs where the external voltage @xmath51 is applied and the constriction is modelled as a film with variable cross - section @xmath52 ( here @xmath53 is the cross - section , @xmath54 is the width and @xmath55 is the thickness of the film ( in this way we model variable thickness microbridge ) . the narrowest part of the film is called the microbridge and the wide parts are assigned as being the leads . the wide part of the film of length @xmath56 is in the normal state and the rest of the film is in the superconducting state ( for relatively small @xmath51 ) . in our calculations we chose @xmath57 and the length @xmath58 was taken large enough to neglect nonequilibrium effects from the ns boundaries located at @xmath59 on the transport properties of the microbridge . the large normal part of the film is needed because in the boundary conditions for @xmath12 and @xmath13 enters the voltage @xmath60 ( see for example ref . @xcite ) and almost all the voltage drop occurs in the normal part of the film , effectively leading to a state with applied constant current even when the microbridge transits to the resistive state . due to the large ratio @xmath61 the transport current does not destroy superconductivity in the wide part of the film and we mainly study the resistive state in the microbridge . in our calculations we assume that the lateral size of the film is much smaller than the london penetration depth @xmath62 ( or effective penetration depth @xmath63 ) and therefore we may neglect screening effects . furthermore , we suppose that the current density is distributed uniformly over the width of the film . this assumption is strictly speaking not valid in the region where the current transits from the wide part to the narrow one ( and vice versus ) . but because the current density is much smaller in the wide part than in the narrow part we may neglect such type of effects ( at least for variable thickness microbridges ) and consider variables which are averaged over the width and thickness of the film . because the length of the microbridge is much larger than its width we may assume an uniform current distribution over the width of the microbridge . thus our system is quasi 1d and we only should take into account the continuity of @xmath64 , @xmath65 , @xmath13 , @xmath32 and conservation of the superconducting ( eq . 6(a ) below ) , normal ( eq . 6(d ) ) , energy ( eq . 6(c ) ) and spectral ( eq . 6(b ) ) currents at the points @xmath66 through the boundary conditions @xmath67 it is easy to find the solution to eqs . ( 1 - 5 ) in the normal region and therefore we only need to solve them in the superconducting region ( in the interval @xmath68 ) with the following boundary conditions : @xmath69 , @xmath70 , @xmath71 , @xmath72 , where @xmath73 and the current density @xmath74 in the points @xmath75 could be found from eq . ( 4 ) . . the results were found from numerical solution of eqs . ( 1 - 5 ) in the presence of coupling terms in eqs . 1(a , b ) . the geometrical parameters of the system are : @xmath76 , @xmath77 , @xmath78 and @xmath79.,scaledwidth=45.0% ] in our numerical calculations we use dimensionless units . the order parameter is scaled by @xmath80 ( @xmath81 is the zero temperature order parameter value in the weak coupling limit ) , distance is in units of the zero temperature coherence length @xmath82 , time in units of @xmath83 and temperature in units of the critical temperature @xmath3 . the current is scaled in units of @xmath84 , the superfluid velocity in units of @xmath85 and the electrostatic potential is in units of @xmath86 . it is useful to introduce the dimensionless inelastic relaxation time @xmath87 which is the main control parameter in the model described by eqs . 1(a , b ) . we used the implicit method for the numerical solution of eqs . ( 1 - 3 ) . the coordinate step of the grid was equal to @xmath88 ( which is much smaller than @xmath2 for the considered temperature interval @xmath89 ) and the time step varied from 0.5 @xmath90 up to 2 @xmath90 depending on the temperature . in our numerical procedure we use an even number of grid points and consequently the center of the microbridge where the order parameter goes to zero ( the phase slip center ) is situated between grid points . therefore any quantity in the phase slip center ( psc ) is in fact calculated at a distance @xmath91 from the psc . for the geometrical parameters of the film we used the following values : @xmath92 , @xmath93 and @xmath58 was varied from 15 @xmath88 ( at @xmath94 ) up to 45 @xmath88 ( at @xmath95 ) . in our calculations we used @xmath96 which covers typical values for many low temperature superconductors ( for example @xcite in nb @xmath97 and in al @xmath98 ) . in the wide part of the film we used @xmath99 which allows us to decrease @xmath100 and the length @xmath58 in order to optimize the calculation time . to find the current - voltage characteristics of the superconducting microbridge we applied a large voltage @xmath101 to the normal reservoirs which induces a large current ( @xmath102 ) resulting in a resistive state in the microbridge . than we decrease @xmath51 in a step - wise manner and we find the time averaged difference of the electrostatic potentials between the ends of the microbridge ( @xmath103 ) as function of @xmath51 . in a similar way it is easy to find the critical voltage @xmath104 ( or critical current @xmath105 ) of the superconducting microbridge when the superconducting state becomes unstable ( by increasing @xmath51 from small values ) . as a result we find the dependence @xmath106 which practically coincides with the dependence @xmath107 . in fig . 2 we plot the time - dependence of the current in the microbridge which illustrates that indeed @xmath108 varies weakly in time and the constant @xmath51 induces an almost constant @xmath108 in our model system even when the microbridge is in the resistive state . at @xmath109 calculated for different values of the inelastic relaxation time in the absence ( a ) and the presence ( b ) of coupling terms in eqs . 1(a , b ) . solid curve is the iv characteristic of a short microbridge in the absence of nonequilibrium effects ( see for example @xcite).,scaledwidth=50.0% ] in fig . 3 we show the current - voltage characteristics of a short microbridge with length @xmath110 at @xmath109 calculated for different values of the inelastic relaxation time @xmath5 in the regime of decreasing applied current ( the length and temperature were chosen close to the parameters of the experiment of ref . @xcite and for tin @xmath111 ) . firstly , we should note the strong influence of the coupling terms in eqs . 1(a , b ) on the iv characteristics : voltage increases ( decreases ) with increasing @xmath5 at @xmath112 in case of the absence ( presence ) of coupling terms . secondly , in both cases the iv curves are hyperbolic - like with no foot-like feature and with hysteresis for relatively large @xmath5 ( the transition to the superconducting state occurs at the retrapping current @xmath113 which could be smaller than the critical current @xmath105 for the transition from the superconducting to the resistive state ) . we plot in fig . 4 the energy dependence of @xmath32 averaged over one oscillation period @xmath114 taken in the phase slip center . the effect of the coupling terms are clearly visible by comparing fig . 4(a ) and 4(b ) . the @xmath115 is on the average negative in case the coupling terms are absent and positive when they are included . in the center of the short microbridge ( i.e. in the phase slip center ) in the absence ( a ) and the presence ( b ) of the coupling terms in eqs . 1(a , b ) . in the inset of fig . 4(b ) we show the distribution of the time averaged @xmath39 and @xmath0 along the microbridge and the leads ( schematically shown).,scaledwidth=45.0% ] let us now discuss the origin of the sign of @xmath116 for the different energies . first we consider the case when the coupling terms are neglected . let us simplify eq . 1(a ) to @xmath117 where we omit the diffusive and coupling terms . if we average eq . ( 7 ) over @xmath114 then we obtain latexmath:[\[%5 \overline{\delta f}_l=-\tau_{in}\frac{\partial f_l^0}{\partial \epsilon}\int_{0}^{t_{\|\delta\|}}\frac{r_2}{n_1}\frac{\partial are solely a function of @xmath0 then @xmath119 . but in general the spectral functions @xmath25 and @xmath23 are a function of two variables @xmath0 , @xmath120 ( at fixed x and @xmath7 - see eq . ( 2 ) ) . to get insight how the ratio @xmath121 changes with varying @xmath120 at _ fixed _ @xmath0 one may solve eq . ( 2 ) with zero second derivative and find that the ratio @xmath121 decreases when latexmath:[$\epsilon \gtrsim for large @xmath120 . in fig . 5 we present the time dependence of @xmath0 , @xmath1 and @xmath39 in the absence of the coupling terms in eqs . 1(a , b ) . first of all we should note that the amplitude of the oscillations of @xmath0 is smaller than @xmath123 . the reason for this effect is the large characteristic time scale for the variation of @xmath0 in comparison with the one of @xmath8 . indeed it is known that nonequilibrium effects may considerably slow down the dynamics of @xmath0 @xcite . when @xmath0 increases ( decreases ) its value is smaller ( larger ) than one could expect from the static dependence @xmath124 ( see for example eq . ( 1.3 ) in @xcite ) for the given value of @xmath125 . for example when @xmath8 reaches zero ( which corresponds to the moment in time when @xmath126 after the phase slip event in fig . 5 ) the order parameter is still much smaller than @xmath123 . due to the time delay in the variation of @xmath0 the supervelocity @xmath127 is different for the same values of @xmath0 taken at different times during the period @xmath114 ( see dashed lines in fig . therefore the ratio @xmath121 in the increasing region of @xmath0 is larger ( smaller ) for @xmath128(@xmath129 in the phase slip center for @xmath130 and @xmath131 for @xmath132 ( see fig . the time averaged potential @xmath39 is positive in the microbridge ( see inset in fig . 4(b ) ) which implies heating of quasiparticles and suppression of the superconducting properties . the larger @xmath5 the stronger the deviation form equilibrium ( see eq . ( 8) and figs . 4(a ) ) and the larger the hysteresis ( see fig . 1(a ) ) . , the supervelocity q and the potential @xmath39 in the center of the short microbridge ( parameters are the same as in fig . 4 ) at @xmath133 and @xmath134 as found from a numerical solution of eqs . ( 1 - 3 ) without coupling terms . the horizontal dashed line marks @xmath135 to illustrate the corresponding different q values at two moments in time.,scaledwidth=50.0% ] consider now the effect of the coupling terms in eqs . 1(a , b ) . the term @xmath136 in eq . 1(a ) may be considered as an additional source of nonequilibrium which is nonzero in the energy interval @xmath137 near local @xmath138 ( in the spatially uniform case with @xmath126 and @xmath139 one has @xmath140 @xmath141 ) - see for example ref . @xcite ) . from eq . 1(b ) we may roughly estimate @xmath142 and the sign of the coupling term @xmath143 is defined mainly by the product @xmath144 ( for discussion of the effect of the coupling terms see also ref . @xcite ) . in the phase slip center and near the ends of the short superconducting microbridge during @xmath114 as found from a numerical solution of eqs . ( 1 - 3 ) including the coupling terms . , scaledwidth=50.0% ] our calculations show ( see fig . 6 ) that near the superconducting leads the product @xmath144 is mainly ( in time ) positive and in the phase slip center it can be both positive and negative ( due to the sign change of q - see fig . 5 ) . it results in positive @xmath145 at @xmath146 - see fig . 4(b ) ( at these energies the ends of the microbridge are determinative because near the leads @xmath0 has a low oscillation amplitude and @xmath147 ) . at lower energies @xmath148 the main source of nonequilibrium is situated in the center of the microbridge where both @xmath144 and @xmath149 changes sign during the oscillations and @xmath150 is negative except at low energies @xmath151 ( see fig . 4(b ) ) . the net effect is a negative time averaged @xmath152 ( see inset in fig . 4(b ) ) and therefore an enhancement of the superconducting properties which explains the decrease of the voltage at fixed current @xmath112 with increase of @xmath5 ( see fig . 3(b ) ) . but at low voltages the effect of the coupling terms ( @xmath153 ) becomes smaller and the role of heating increases . this subsequently leads to an increase of the voltage at fixed current @xmath154 with increasing @xmath5 and to an hysteresis of the iv curves ( but which is smaller than in the case without coupling terms - compare figs . 5(a ) and 5(b ) ) . the origin of the here predicted time averaged heating of quasiparticles is different from joule dissipation . indeed , the source of nonequilibrium @xmath155 changes sign during oscillations of the order parameter while the term @xmath156 is always positive . time averaged heating appears only due to the dependence of the spectral functions both on @xmath0 and @xmath1 and due to the difference in time variation of @xmath0 and @xmath8 . energy for heating of the quasiparticles ( and cooling due to the term @xmath157 in the kinetic equation ) comes from the energy @xmath156 delivered by the external source of the current ( this energy goes also for example to the heating of phonons due to presence of the relaxation term in the kinetic equations ) . our result is opposite to the one of refs . @xcite where a time averaged cooling of quasiparticles and an enhancement of superconductivity was found at high voltages ( when @xmath158 ) in the absence of the coupling terms in eqs . 1(a , b ) . the differences with our calculations are that we take into account : i ) the backaction of the nonequilibrium quasiparticle distribution function on the dynamics of @xmath0 ; and ii ) the dependence of the spectral functions @xmath23 , @xmath24 and @xmath25 on _ both _ @xmath0 and @xmath120 . there is also one more difference with refs . @xcite - we did not neglect the relaxation term in eq . ( 7 ) at high voltages ( see eq . ( 3 ) in ref . @xcite and eq . ( 16 ) in ref . @xcite ) . by omitting this term one may obtain cooling with @xmath159 if one chooses the initial condition @xmath160 when @xmath161 in the center of the microbridge ( and when the spectral functions @xmath25 and @xmath23 depend only on @xmath0 as was assumed in refs . @xcite ) . but this choice of the initial condition is not obvious . by choosing a different initial condition ( large negative @xmath32 at @xmath161 ) one may obtain heating in this case too . only the presence of the relaxation term in eq . ( 7 ) resolves this problem , because it defines undoubtedly the time averaged value of @xmath32 ( see eq . note that due to model assumptions for spectral functions of refs . @xcite @xmath162 has to be equal to zero at any voltage if one does not neglect the relaxation term . from a physical point of view the relaxation term in eq . ( 7 ) tells us that the accumulation of the heat/cold ( after sudden change of the current for example ) occurs on a time scale which is proportional to @xmath163 and one should take it into account to find the true value of @xmath162 and @xmath114 corresponding to the given value of the current . it provides a strong dependence of the iv characteristics and @xmath164 on @xmath5 even at high voltages ( see figs . 3(a ) and 4(a ) ) while in refs . @xcite no dependence on @xmath5 was found in this limit . according to our numerical calculations the iv characteristics of short dirty superconducting microbridges are hyperbolic - like and hysteretic ( at relatively large @xmath5 and/or low temperatures ) in the temperature interval @xmath165 ( calculated and not presented here the iv characteristics at @xmath166=0.92 - 0.99 are qualitatively similar to those of figs . 3(a , b ) ) . we believe that the experimentally found foot-like features in the iv characteristics of short superconducting microbridges is connected with the geometry of the studied microbridges - they were short and _ wide _ ( in experiments @xcite the width @xmath167 ) . for such a geometry the current density distribution over the width of the microbridge ( constriction ) should be nonuniform with sharp peaks near the edges of the microbridge @xcite . in ref . @xcite it was shown that a nonuniform current density distribution in the superconductor may lead to slow vortex motion at low currents and phase slip lines at high currents @xcite and to an iv characteristic ( see for example fig . 4 in ref . @xcite ) which resembles the experimental results of refs . moreover the sharp transition from low to high voltages occurs at @xmath168 which is inversely proportional to @xmath5 @xcite - the same result was found in the experiment on tin superconducting microbridges @xcite . therefore , we predict that the iv characteristics of short ( @xmath11 ) superconducting microbridges should change from hyperbolic - like ( in case when @xmath169 ) to the one with a well - pronounced foot-like structure at low voltages when @xmath170 . good candidates are sn , in , pb and al with relatively large coherence length ( @xmath171 at @xmath172 ) . we expect that our results will be tested in the near future because recently it has become possible to fabricate microbridges with length @xmath173 500 nm and width @xmath174 nm ( see for example ref . @xcite ) which satisfy the condition @xmath175 at @xmath176 . this work was supported by the russian foundation for basic research , russian agency of education under the federal target programme `` scientific and educational personnel of innovative russia in 2009 - 2013 '' , flemish science foundation ( fwo - vl ) and the belgian science policy ( iap ) .	we predict heating of quasiparticles driven by order parameter oscillations in the resistive state of short superconducting microbridges . the finite relaxation time of the magnitude of the order parameter @xmath0 and the dependence of the spectral functions both on @xmath0 and the supervelocity @xmath1 are the origin of this effect . our result is opposite to those of aslamazov and larkin ( zh . eks . . fiz . * 70 * , 1340 ( 1976 ) ) and schmid , schn and tinkham ( phys . rev . b * 21 * 5076 ( 1980 ) ) where cooling of quasiparticles was found .
a key observable statistic that describes the evolution of the galaxy population is the average sfr as a function of epoch and of galaxy stellar mass . over the last decade , much progress has been made towards delineating , and in part understanding , the relationship between star formation history and stellar mass . `` archaeological '' studies of present - day galaxies demonstrate a strong correlation between star formation history and present - day stellar mass in the sense that the bulk of stars now in massive galaxies must have formed at earlier epochs than stars now in less massive galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? surveys of cosmologically distant galaxies have revealed a related trend , often referred to as `` downsizing '' in that context . downsizing most commonly refers to the observation that intense star formation ( @xmath7 ) becomes increasingly limited to systems of lower and lower mass as redshift decreases @xcite . recently , this phenomenology has been recast in more quantitative terms through study of the specific sfr ( i.e. , sfr per unit stellar mass ) as a function of redshift , finding that at @xmath8 , low - mass galaxies appear to be forming stars at a higher rate per unit mass than their more massive cousins @xcite . there are two important parts of this empirical picture that require clarification . firstly , many studies use sfr indicators which are highly susceptible to dust extinction ( e.g. , [ oii ] or uv - optical sed - derived sfrs ; @xcite ) . secondly , all studies to date have relied on individual detections to allow sfr derivation . neglecting sfr contributions from galaxies with individually undetectable star formation tracers biases the census of sfr at all masses . however , the bias is especially severe at low masses where the sfr limiting sensitivity implies that all low - mass galaxies with individual sfr detections have extremely high specific sfr . in this letter , we present an analysis of spitzer 24@xmath0 observations of two combo-17 fields designed to circumvent these two key limitations . we estimate sfr by combining indicators of obscured and unobscured star formation , and attempt to derive average sfrs through stacking of galaxy sub - samples with individually - undetected sfr indicators . in 2 , we summarize the data and galaxy samples . in 3 , we describe our sfr measurements . in 4 , we present the contribution of different bins in stellar mass to the cosmic sfr , and present the evolution of specific sfr with lookback time , again as a function of stellar mass . we discuss our results in 5 . we adopt a cosmology with h@xmath9=70kms@xmath10mpc@xmath10 , @xmath11=0.3 and @xmath12=0.7 . the combo-17 survey @xcite has imaged three @xmath13 fields in five broad and 12 intermediate optical bands from 350 to 930 nm with the goal of deriving high - quality photometric redshifts at the median galaxy magnitude of @xmath14 @xcite . ] for @xmath15 galaxies with @xmath16mag in each field . the rest - frame luminosities in the @xmath17 , @xmath18 and @xmath19 band , and a synthetic uv2800 band centered at 280 nm have been derived for all combo-17 galaxies with photo - z s . by fitting the combo-17 multi - band photometry with a library of sed templates , @xcite estimated the galaxies stellar mass - to - light ratio ( @xmath20 ) , and hence stellar mass for all @xmath16 galaxies based on a @xcite initial mass function ( imf ; such an imf has stellar masses similar to within 10% compared to those derived using a @xcite or @xcite imf ) . from these data , @xcite constructed stellar mass functions in four even redshift slices between @xmath21 and @xmath22 . two of three combo-17 fields , the extended chandra deep field south ( cdfs ) and the abell 901/902 ( a901 ) have been observed by the multiband imaging photometer on _ spitzer _ ( mips : @xcite ) at 24@xmath0 . the reduced 24@xmath0 images cover an area of @xmath23 around the cdfs @xcite and of @xmath24 in the a901 field @xcite . both images have a pixel scale of @xmath25pixel@xmath10 , a point spread function ( psf ) with full width at half maximum ( fwhm ) of @xmath266@xmath27 . the iraf / daophot package and empirically - determined psfs are used to simultaneously fit multiple sources to the 24@xmath0 images and derive total flux for each detected object to a depth of 83@xmath28jy at the 5@xmath29 level ( to a completeness of 80% ; see also @xcite for details of data reduction , source detection and photometry ) . to match these 24@xmath0 catalogs to the combo-17 galaxy samples , a position tolerance of @xmath30 is adopted . if more than one galaxy exist within the tolerance , the nearest one is selected . the overlap area between combo-17 and mips imaging contains 9785 and 12995 galaxies at @xmath31 in the cdfs ( @xmath1800 square arcminutes ) and the a901 field ( @xmath1850 square arcminutes ) , respectively . of them , 1735 ( 1872 ) are individually detected at 24@xmath0 . only @xmath34% ( 6% ) of the 24@xmath0-detected sources , namely 64 ( 108 ) , have multiple ( mostly two ) combo-17 galaxies within the tolerance . we select galaxies with stellar mass @xmath32m@xmath33 in the redshift range @xmath2 , resulting in a sample of 6893 ( 8184 ) galaxies in the cdfs ( a901 ) field . of the 6893 ( 8184 ) galaxies , 1201 ( 1250 ) are individually detected at 24@xmath0 . the _ average _ 24@xmath0 flux of the galaxies individually undetected by mips will be estimated by stacking , as described in the next section . the samples are complete at mass limit @xmath11 , 1.8 , 3 , 6@xmath34m@xmath33 for redshift @xmath350.3 , 0.5 , 0.7 , 0.9 @xcite . owing to mismatches in depth between wide - area ir surveys ( i.e. , iras ) and well - characterized optical galaxy surveys ( i.e. , the sdss and 2dfgrs ) , it is challenging to produce an appropriate and well - defined local comparison sample . we have attempted to produce an adequate local control sample by collecting a sample of 2177 galaxies from the nasa / ipac extragalactic database ( ned ) with 2mass @xmath36-band magnitude @xmath37 in the volume of @xmath38 and galactic latitude @xmath39 . assuming a @xmath36-band @xmath20 of 0.6m@xmath33/l@xmath33 and a kroupa imf , stellar masses were derived from the @xmath36-band absolute magnitude . the sfr is estimated from the total ir luminosity , as determined from iras observations . the local sample is not a complete volume - limited sample . however , we estimate that above @xmath40m@xmath33 the sample is @xmath165% complete with an incompleteness which is largely geometric in origin ( because some areas of the sky were not spectroscopically surveyed as thoroughly as others ) . if indeed the incompleteness is primarily geometric in origin , the average sfrs at a given stellar mass should not be biased . and fig . [ ssfr ] shows that none of our conclusions actually depend on the @xmath41 data ; the purpose of its inclusion is simply to extend the trends to @xmath41 . ] see @xcite for more details about the local sample selection and completeness . cccc @xmath10.01 & ... & @xmath42 & @xmath43 + 0.2 0.4 & @xmath44 & @xmath45 & @xmath46 + 0.4 0.6 & @xmath47 & @xmath48 & @xmath49 + 0.5 0.8 & @xmath50 & @xmath51 & @xmath52 + 0.8 1.0 & @xmath53 & @xmath54 & @xmath55 + we split the sample galaxies into four redshift slices evenly covering the range from @xmath21 to @xmath22 . in each redshift slice , sample galaxies are divided into five stellar mass bins : four bins with a bin width of 0.5dex spanning from @xmath56m@xmath33 to @xmath57m@xmath33 , and one bin for @xmath58m@xmath33 . each sub - sample of galaxies defined by optical data contains individually - detected and individually - undetected objects at 24@xmath0 . following @xcite we estimated the total 24@xmath0 flux for the sub - sample of individually - undetected objects through stacking . the total 24@xmath0 flux of the sub - sample is then obtained by adding the individual fluxes from the detected objects and stacked flux from the individually - undetected objects . our sfr indicator , the total ir luminosity ( 8 - 1000@xmath0 ) , is extrapolated from the 24@xmath0 luminosity using three sets of local ir sed templates from @xcite , @xcite and @xcite . in @xcite , we use stacking of ir - bright galaxies with @xmath59 at 70@xmath0 and 160@xmath0 to show that the average ir sed of an ir - bright galaxy is spanned by these local template sets , lending credibility to the total ir luminosity estimated here to within a factor of two . we use the bolometric luminosity produced by young stars ( uv+ir ) to estimate the sfr , a procedure that accounts for both obscured and unobscured star formation . the total uv luminosity ( 1216-3000 ) is estimated from the combo-17 2800 rest - frame luminosity @xmath60 as @xmath61=1.5@xmath62 . we used the conversion from @xcite to convert the uv+ir into sfr . the sfr is calibrated to a @xcite imf . using the sum of stellar masses for each sub - sample from @xcite , we calculated the average specific sfr , i.e. @xmath63 , for all redshift and stellar mass bins . formal uncertainties were estimated from bootstrapping . for each redshift and stellar mass bin , the average between the cdfs and a901 was taken as the final result ; the difference between the two is taken as a measure of field - to - field variance and included into the uncertainties . the estimates of both sfrs and stellar masses are affected by systematic errors , that other studies have estimated to be @xmath64dex . these systematic uncertainties are not included in our error bars , but are substantially smaller than the dynamic range of the trends discussed in this letter . due to the combo-17 @xmath65-band selection ( @xmath16 ) , our sample galaxies are incomplete for low - mass bins at @xmath66 , where low - mass red galaxies with little blue radiation from newly - formed stars fail to make the @xmath65-band selection cut but low - mass blue galaxies containing more on - going star formation remain available for the selection . the selection effect may lead to a potential overestimate of the specific sfr for these incomplete bins . the specific sfr at @xmath67 was estimated using the local sample only for three high - mass bins where the sample is representative . , despite the factor @xmath17 decline in the total sfr in that interval . , scaledwidth=48.0% ] in the previous section , we have determined the average sfrs for galaxies split into stellar mass and redshift bins . in order to determine the contribution of different stellar mass bins to the cosmic ( average ) sfr , we must combine the average sfrs at a given mass with the space density of galaxies in that same mass range , as a function of redshift . for this purpose , we adopt the stellar mass functions of @xcite for @xmath69 , and @xcite for @xmath41 , although the results are unaffected to within the uncertainties if we adopt stellar mass functions from , e.g. , @xcite , @xcite or @xcite instead . we calculate the volume - averaged sfr over the redshift range @xmath70 for three galaxy stellar mass bins : @xmath71 , @xmath72 and @xmath73 . the stellar mass functions from @xcite are derived from @xmath125000 combo-17 galaxies selected from three separated @xmath74 fields . therefore , the field - to - field variance is somewhat suppressed in this way ( as one can see from comparison of the total sfr , derived from two combo-17 fields by @xcite compared to the smooth trends followed by our mass - binned cosmic sfrs ) . the local data point is not available for the stellar mass bin @xmath75 . the results are listed in table [ table ] and shown in fig . [ sfrds ] , along with current measurements of the cosmic sfr density available in the literature @xcite . @xcite constructed sfr functions using the same data sets and sfr estimator ( uv+ir ) as adopted here . the data points given by their sfr functions are shown also in this plot . our estimate of the cosmic sfr density , i.e. the sum of the volume - averaged sfrs from the three stellar mass bins ( contribution from galaxies of stellar mass @xmath76m@xmath33 is negligible ) , is in good agreement with bell et al.s and other measurements from the literature . [ sfrds ] shows that the volume - averaged sfr of galaxies in each of the three stellar mass bins decreases towards the present day with an indistinguishably similar slope since @xmath22 . in other words , the contributions to the cosmic sfr density from galaxies of different stellar mass ranges has not changed with cosmic epoch during the last 8gyr . galaxies in the @xmath77 contain the bulk of the overall star formation at all cosmic epochs which we have examined . these galaxies also dominate the cosmic stellar mass density @xcite . a smaller part of the overall star formation takes place in galaxies with @xmath78 . relatively little star formation is associated with massive galaxies with @xmath79m@xmath33 . , as function of redshift for galaxy samples of differing stellar masses . the average sfr is derived from the average bolometric luminosity ( uv+ir ) for galaxy subsets limited in mass and redshift . the data points in the hatched region are incomplete due to the combo-17 source selection ( @xmath16 ) . the gray strip illuminates the overall specific sfr , i.e. the ratio of the cosmic sfr density to the cosmic stellar mass density , following @xcite and @xcite . local data points are estimated for three high - mass bins using a local galaxy sample with near - ir and iras observations . the errorbar represents the typical uncertainty of our specific sfr determinations . the dark - gray dashed line is the specific sfr required to double the mass of a galaxy by the present day . data points above the mass - doubling line represent galaxies in intense star formation mode and those below the line represent galaxies in quiescent star formation mode . for all galaxy sub - samples of different stellar masses , the specific sfr increases with redshift out to @xmath6 . at all redshifts , more massive galaxies always have lower specific sfr than less massive galaxies . these relations delineate a picture in which lower - mass galaxies transition from an intense into a quiescent star formation phase since @xmath80 , whereas most massive galaxies have formed stars quiescently across this epoch . open symbols show results from feulner et al . ( @xcite ) based on sfr indicator rest - frame uv luminosity for stellar mass bins ( converted to a kroupa imf ) @xmath81 ( _ triangles _ ) , @xmath82 ( _ squares _ ) and @xmath83 ( _ circles _ ) . , scaledwidth=48.0% ] fig . [ ssfr ] presents our measurements of specific sfr , i.e. sfr per unit stellar mass . data points of the same stellar mass bin are linked together , showing the redshift evolution of the specific sfr . data points of the same redshift delineate the specific sfr as a function of stellar mass . those enclosed in the shaded region are in bins affected by serious incompleteness due to the combo-17 flux limit . for comparison , we overplotted the average specific sfr for all galaxies , i.e. the ratio of the cosmic sfr density to the cosmic stellar mass density , calculated using @xcite and @xcite . as shown in fig . [ ssfr ] , the average specific sfr for a galaxy stellar mass bin increases rapidly with redshift , in qualitative agreement with previous studies ( e.g. , * ? ? ? * ; * ? ? ? compared to our results , sfr estimates from rest - frame uv luminosity ( open symbols in fig . [ ssfr ] ; * ? ? ? * ) produce lower specific sfrs for high - mass galaxies and higher specific sfrs for low - mass galaxies due to the reasons mentioned in [ intro ] . yet , the change in specific sfr is nearly equally strong for all galaxy mass bins . at all redshifts , the population - averaged specific sfr of less massive galaxies is higher than that of more massive galaxies . note that , as expected , galaxies of stellar mass around @xmath84 have specific sfr comparable to the ` cosmic average ' specific sfr , as they contribute most to both the cosmic average sfr and stellar mass budget . we combine _ spitzer _ 24@xmath0 observations with combo-17 data to explore the evidence for the ` downsizing ' of the star formation activity since @xmath6 . specifically , we estimate sfr from the uv ( for unobscured star formation ) and the ir luminosities ( for obscured star formation ) for a sample of @xmath115000 galaxies selected from the cdfs and a901 fields . through image stacking , we account for the sfr contributions of otherwise undetectable galaxies . our measurements of average sfr for galaxy subsets limited in mass and redshift account for contributions from both star - forming galaxies and quiescent galaxies in each subset . we find that the relative drop in sfr from @xmath80 to the present day is nearly identical for different sub - samples that span a range of @xmath1100 in stellar mass . that is , all galaxy samples , independent of mass , exhibit the same gradual decline of star formation ( and specific star - formation ) over the last 8 gyr . at all redshifts , we find that high - mass galaxies have substantially lower specific sfr than their low - mass cousins . while many of the qualitative features of this paper have been understood for some time ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , this letter quantifies the relationships between specific sfr , galaxy stellar mass and redshift . ` archaeological ' downsizing that the stars in massive galaxies formed at earlier times than the stars in low - mass galaxies is clearly recovered in this paper , in fig . on average , low - mass galaxies have specific sfrs capable of more than doubling their stellar mass over the hubble time ( at all redshifts ) , whereas galaxies with @xmath79m@xmath85 are not capable of such growth at @xmath86 from _ in situ _ star formation alone . arguably the most striking and new contribution of this letter is that the decline in sfr from @xmath80 to the present occurs at the same rate in galaxies of all masses . a number of mechanisms have been postulated to drive the demise of star formation since @xmath80 : agn feedback , the declining galaxy merger rate , gas consumption , changes in the availability of cooling or infalling fresh gas , and a host of other possibilities . we speculate that the declining supply of new gas falling in from the cosmic web at @xmath8 seems an attractive candidate for driving much of the phenomenology that we observe @xcite . the roughly constant rate of decline of sfr and ssfr for galaxies of widely differing masses is in contrast to the predictions of theoretical models in which cooling and hence star formation is quenched by heating due to radio - loud agn ( ( e.g. , * ? ? ? * ; * ? ? ? * ) ; but see also @xcite ) . this process is more important in more massive galaxies for two reasons . first , more massive galaxies are more likely to host massive black holes , and more massive black holes are generally assumed to be capable of producing more energy and stronger heating . second , it is generally assumed in such models that agn can only heat gas that is cooling from a quasi - hydrostatic hot halo , which form in large mass halos ( @xmath8710@xmath88m@xmath33 ) as opposed to the `` cold mode '' flows that are dominant in lower mass halos . for example , the models of @xcite predict a decline of a factor of @xmath18 from @xmath6 to the present , in the sfr contributed by galaxies more massive than @xmath89m@xmath33 ( see their figure 8) , while the sfr contributed by less massive galaxies declines by a factor of about 24 over this same interval . we find , using an independent model with radio - mode heating by agn ( somerville et al . in preparation ) , that the specific star formation rate in galaxies with stellar mass greater than @xmath57m@xmath33 has declined by a factor of @xmath115 since @xmath6 , while the specific sfr in the lower mass bins ( the same bins used here ) , declines by only a factor of 35 . the precise rate of decline of the sfr as a function of galaxy mass is no doubt sensitive to the details of the physics of star formation , but the prediction of a strongly differential ` quenching ' of star formation by agn feedback in the form of radio heating , as it is currently implemented , seems inevitable . our results therefore suggest that radio mode heating by agn is not the dominant mechanism responsible for the decrease in cosmic sfr from @xmath6 to the present . e. f. b. was supported by the emmy noether programme of the deutsche forschungsgemeinschaft . c. w. was supported by a pparc advanced fellowship . we thank the anonymous referee for helpful suggestions that improved our manuscript . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . bauer , a. e. , drory , n. , hill , g. j. , & feulner , g. 2005 , apj , 621 , l89 bell , e. f. , mcintosh , d. , katz , n. , & weinberg , m. d. 2003 , apjs , 149 , 289 bell , e. f. , et al . 2005 , apj , 625 , 23 bell , e. f. , zheng , x. z. , papovich , c. , borch , a. , wolf , c. , & meisenheimer , k. 2007 , apj , submitted borch , a. , et al . 2006 , a&a , 453 , 869 bower , r. g. , et al . 2006 , mnras , 370 , 645 brinchmann , j. , & ellis , r. s. 2000 , apj , 536 , l77 brinchmann , j. , et al . 2004 , mnras , 351 , 1151 bundy , k. , et al . 2006 , apj , 651 , 120 chabrier , g. 2003 , apj , 586 , l133 chapman , s. c. , blain , a. w. , ivison , r. j. , & smail , i. r. 2003 , nature , 422 , 695 chary , r. , & elbaz , d. 2001 , apj , 556 , 562 cowie , l. l. , songaila , a. , & hu , e. m. 1996 , aj , 112 , 839 croton , d. j. , et al . 2006 , mnras , 365 , 11 daddi , e. , et al . 2005 , apj , 631 , l13 dale , d. a. , & helou , g. 2002 , apj , 576 , 159 drory , n. , et al . 2004 , apj , 608 , 742 feulner , g. , gabasch , a. , salvato , m. , drory , n. , hopp , u. , & bender , r. 2005 , apj , 633 , l9 fontanta , a. , et al . 2006 , a&a , 2006 , 459 , 745 gallazzi , a. , charlot , s. , brinchmann , j. white , s. d. m. , & tremonti , c. a. 2005 , mnras , 362 , 41 hopkins , a. m. , 2004 , apj , 615 , 219 lagache , g. , et al . 2004 , apjs , 154 , 112 kauffmann , g. , et al . 2003 , mnras , 341 , 54 kroupa , p. 2001 , mnras , 322 , 231 kroupa , p. , tout , c. a. , & gilmore , g. 1993 , mnras , 262 , 545 juneau , s. , et al . 2005 , apj , 619 , l135 neistein , e. , van den bosch , f. c. , & dekel , a. 2006 , mnras , 372 , 933 noeske , k. g. , et al . 2007 , apj , in press ( astro - ph/0701924 ) panter , b. , jimenez , r. , heavens , a. f. , & charlot , s. astro - ph/0608531 papovich , c. , et al . 2004 , apjs , 154 , 70 papovich , c. , et al . 2006 , apj , 640 , 92 rieke , g. h. , et al . 2004 , apjs , 154 , 25 thomas , d. , maraston , c. , bender , r. , & de oliveira , c. m. 2005 , apj , 621 , 673 wolf , c. , et al . 2003 , a&a , 401 , 73 wolf , c. , et al . 2004 , a&a , 421 , 913 zheng , x. z. , bell , e. f. , rix , h .- w . , papovich , c. , le floch , e. , rieke , g. h. , & prez - gonzlez , p. g. 2006 , apj , 640 , 784 zheng , x. z. , dole , h. , bell , e. f. , le floch , e. , rieke , g. , h. , rix , h .- w . , & schiminovich , d. 2007 , apj , submitted	we combine _ spitzer _ 24@xmath0 observations with data from the combo-17 survey for @xmath115,000 @xmath2 galaxies to determine how the average star formation rates ( sfr ) have evolved for galaxy sub - populations of different stellar masses . in the determination of @xmath3sfr@xmath4 we consider both the ultraviolet ( uv ) and the infrared ( ir ) luminosities , and account for the contributions of galaxies that are individually undetected at 24@xmath0 through image stacking . for all redshifts we find that higher - mass galaxies have substantially lower specific sfr , @xmath3sfr@xmath4/@xmath3m@xmath5@xmath4 , than lower - mass ones . however , we find the striking result that the rate of decline in cosmic sfr with redshift is nearly the same for massive and low - mass galaxies , i.e. _ not _ a strong function of stellar mass . this analysis confirms one version of what has been referred to as ` downsizing ' , namely that the epoch of major mass build - up in massive galaxies is substantially earlier than the epoch of mass build - up in low - mass galaxies . yet it shows that star formation activity is _ not _ becoming increasingly limited to low - mass galaxies towards the present epoch . we argue that this suggests that heating by agn - powered radio jets is not the dominant mechanism responsible for the decline in cosmic sfr since @xmath6 , which is borne out by comparison with semi - analytic models that include this effect .
as is well known , albert einstein was convinced that god does nt play dice with the universe . to that , niels bohr used to reply to stop telling god what s / he has to do with her / his dice . g. cramer , in more recent times , added to this the following @xcite : `` if ` god plays dice , ' as einstein ( 1932 ) has declined to believe , one would at least like a glimpse of the gaming apparatus that is in use . '' having a glimpse of the apparatus means to _ understand _ the nature of the game that is played behind the quantum scenes , and this in turn means _ to understand quantum mechanics _ , something which most physicists believe is a hopeless goal , in accordance with richard feynman celebrated admonition @xcite : `` [ ... ] that nobody understands quantum mechanics . '' but what does it mean to understand quantum mechanics ? and , more generally , what does it mean to understand something in science ? different ( more or less sophisticated ) answers are of course possible , depending on the perspective which is adopted . but the majority of scientists will certainly agree that , roughly speaking , to understand in science means to build a theory with a sufficient explanatory power , which can suitably elucidate the observed phenomena and make confirmable predictions . then , why quantum mechanics , which today has reached the level of a fully mathematized theory , and whose predictions have been confirmed experimentally to an extremely high degree of accuracy , is still considered a theory that nobody really understand , as feynman used to proclaim ? again , the answer to this question will vary depending on the physicist that respond , but the majority will certainly say that although quantum theory describes and predicts with accuracy the behavior of quantum entities , the problem is that their behavior has nothing to do with that of classical macroscopic entities of our everyday experience . therefore , we simply lack of concrete models that would allow us to visualize a quantum entity and understand , by analogy , its behavior , since we humans can only visualize models made of concrete objects , belonging to our ordinary spacetime theatre . let us think for instance of an _ armilla _ ( also known as _ spherical astrolabe _ ) : an ancient little model of the objects in the sky , consisting of a spherical framework of graduated metal circles ( `` armilla '' means `` circle '' in latin ) , centered on earth , representing the lines of celestial longitude and latitude , the ecliptic , and other astronomically important features . the armilla , which was invented by the greek astronomer eratosthenes ( 276 -194 bce ) , was used , among other things , as a valid teaching tool , i.e. , as a mean to conveniently visualize the structure and behavior of the astronomical solar system which , otherwise , because of its astronomical size , would have been rather difficult to visualize as a whole , and therefore to understand . being the solar system an astronomical object , it is clear that any model of it will not be a real - size one , but a miniature . and it is precisely because the armilla was a miniature , whose structure could be viewed as a whole , with a glance , that it had such a great explicative power and allowed to really understand the structure and behavior of the solar system ( from an ancient geocentric point of view ) . the above is just to emphasize that the ability to create concrete , easily viewable models , is an important tool of the scientific enterprise , above all with regard to the possibility of fully understanding the functioning of the objects of its study . now , if for astronomical entities models have to be miniatures , when it is about the representation of microscopic entities , models of course have to be larger than the original . as a paradigmatic example , we can consider the many different large scale atomic models that have been used in the course of history , to visualize and explore the atomic hypothesis . let us think for instance of the very simple `` billiard ball '' atomic models of democritus and dalton , the more sophisticated thomson s `` plum pudding '' model , or the rutherford s `` armilla - like '' solar system model , subsequently modified by bohr , by restricting the `` allowed '' orbits and energies for the electronic satellites . this is probably one of the reasons why bohr s atomic model , despite being totally obsolete , is sometimes still taught today as an introduction to quantum mechanics , as it has the merit of being somehow still viewable by the students , who have then the impression of understanding what s going on . but of course , bohr s atomic model is already a `` crazy '' one , and bohr himself was accustomed to say that the real question was to determine whether it was crazy enough to have a chance of being right ! as we today all know , quantum theory , despite its weirdness , has certainly being confirmed to be right . but the price we have paid for this is , apparently , to renounce the possibility of understanding it . indeed , what kind of model can we conceive today that would allow us to understand atoms without invoking magic images ? ( like the one of an elevator which , in the case of bohr s model , would be able to bring an electron to different floors , i.e. , orbits , without ever sojourning in between of them ) . to quote heisenberg @xcite : `` the atom of modern physics can only be symbolized by a partial differential equation in an abstract multidimensional space . only the experiment of an observer forces the atom to indicate a position , a color and a quantity of heat . all the qualities of the atom of modern physics are derived , it has no immediate and direct physical properties at all , i.e. every type of visual conception we might wish to design is , _ eo ipso _ , faulty . '' the same of course can be said about the constituents of the atomic structures , the so - called _ elementary particles _ , which in fact do not behave as particles , and can certainly not be _ understood _ ( i.e. , visualized ) as little tiny corpuscles moving around in the three - dimensional euclidean space @xcite . so , how should we understand the reality of the microscopic quantum entities ? and , more importantly , can we understand it ? the majority of physicists believe that such an understanding is simply impossible , and this because we lack of simple models ( apart from the abstract model of the mathematical theory ) that would guide our intuition and allow us to understand the reasons for the observed quantum weirdness . but , is all this true ? do we really lack of simple concrete models that can help us to understand the behavior of elementary quantum entities ? in fact , and contrary to what is usually believed , these models exist , since many years now , and can certainly be used to convincingly visualize some of the strange behaviors of quantum entities , as they interact with the experimental contexts . they have been invented by the belgian physicist diederik aerts , one of the founders of so - called _ geneva - brussels school on the foundations of physics _ , and according to the present author their importance have been seriously underappreciated by the physics community , despite the role they have played in clarifying many of the conceptual challenges of quantum physics , and in facilitating the development of more advanced conceptual and mathematical formalisms that go beyond the quantum ( hilbertian ) and classical ( phase space ) descriptions . also , apart their interest in sponsoring new innovative approaches to understand the complexity of our reality like for instance the _ hidden - measurement approach _ @xcite and _ creation - discovery view _ @xcite these models can be advantageously used to reveal what could be a possible physical content of the theory , beyond its rather abstract and counterintuitive formalism . in this paper we shall present two of aerts historical quantum machine - models , whose operations are based on simple breakable elastic bands . the first one is called the _ spin quantum machine _ , or _ sphere model _ @xcite , and can easily replicate the quantum probabilities associated with the spin measurement of a spin-@xmath0 entity , for instance in a typical stern - gerlach experiment . the model allows to understand what possibly distinguish quantum from classical probabilities : if the latter correspond to a situation of lack of knowledge about the state of the system , the former may correspond to situations where there is full knowledge of the state of the entity , but maximum lack of knowledge about the interaction between the measurement apparatus and the entity . the second machine - model that we shall describe is called the _ connected vessels of water _ model @xcite . as its name indicates , the original model is constructed using vessels , tubes and water . however , we shall present here an alternative version of it , only using an elastic band , in accordance with the title of the present paper . the model is able to violate bell s inequalities , as coincidence measurements on entangled states can do , and since everything in it happens under our eyes , part of the mystery of so - called `` spooky actions at a distance '' ( i.e. , epr non - local correlations ) will be revealed , and therefore explained : bell s inequalities ( and corresponding bell s locality hypothesis ) can only be violated when , during a coincidence experiment , correlations that were nt present before the experiment are literally created by and during the experiment itself . before describing aerts spin quantum - machine ( sqm ) , let us recall some of the basic properties of a spin-@xmath0 quantum entity . its state @xmath1 belongs to a 2-dimensional hilbert space @xmath2 , and can always be represented as a superposition of `` up '' and `` down '' spin states , relative to an a priori given direction @xmath3 : @xmath4 and @xmath5 are complex numbers obeying the normalization condition @xmath6 . if we denote by @xmath7 the spin vector observable , its components are known to obey the commutation relation @xmath8=i\hbar s_{\hat{z}}$ ] , and the equations : @xmath9 because of the normalization condition , one can always write ( apart from a global phase factor with no physical meaning ) @xmath10 and @xmath11 , so that if @xmath12 and @xmath13 are taken to be the polar angles of a unit vector @xmath14 , then to each state vector @xmath1 one can associate , bijectively , a unit vector @xmath14 , which can be written as @xmath15 in polar coordinates , or as @xmath16 in cartesian coordinates . according to the above one - to - one pauli mapping , all spin states can be visualized as points on a 3-dimensional unit sphere , indexed by a specific unit vector ( this is also known as a poincar sphere representation ; see for instance the discussion in @xcite , pp . more precisely , if @xmath1 is given by ( [ spin state ] ) , and @xmath14 is given by ( [ unit vector ] ) , then we have @xmath17 ( where the symbol `` @xmath18 '' means here `` equals , up to a non zero overall phase factor '' ) , and clearly @xmath19 . to each unit vector @xmath20 ( i.e. , to each point on the unit sphere ) , one can associate the projection operator @xmath21 onto the set of states having spin `` up '' along direction @xmath20 . therefore , according to born rule , the probability that a spin measurement along direction @xmath20 gives the outcome `` up '' ( i.e. , value @xmath22 ) , when the system is prepared in state @xmath23 , is given by : @xmath24 to calculate this probability , one can reason as follows : if @xmath25 is the angle between @xmath14 and @xmath20 , i.e. , the angle that one needs to rotate @xmath14 , along direction @xmath26 , to reach @xmath20 , and if we denote by @xmath27 the corresponding rotation operator , with @xmath28 , we have @xmath29 . then , exploiting the remarkable properties of the exponential function and of the spin @xmath30 matrices ( see any good book of quantum mechanics ) , it is not difficult to show that the rotation operator @xmath31 can be written in the simple form : @xmath32 therefore , considering that the unit vector @xmath33 is orthogonal to @xmath20 , according to ( [ x - y component equation ] ) we have @xmath34 , and one immediately finds that : @xmath35 and of course , since @xmath36 , we also have @xmath37 as is well known , orthodox quantum mechanics does nt explain the reason of probabilities ( [ probability up2 ] ) and ( [ probability down ] ) , i.e. , where the indeterminism they subtend comes from and how the system is able to actualize outcomes that are only potential prior to the measurement . of course , what is lacking in the above description , and more generally in the axiomatic formulation of orthodox quantum mechanics , is the measuring apparatus , i.e. , a description of how the apparatus precisely interacts with the system during an idealized measurement . in the case of a spin measurement , the apparatus is for instance the one used in a typical stern - gerlach experiment , made of a magnetic field with a strong gradient ( exerting a torque on the magnetic moment of the quantum entity ) , plus a detection screen used to reveal which beam deflection ( upward or downward ) has been actualized . unfortunately , a direct observation of what exactly happens `` behind the scenes , '' when the spin quantum entity interacts ( in a non - local way ) with the apparatus and manifests its presence by means of a specific ( upward or downward ) spot on the screen , appears to be beyond our today ( and perhaps also our tomorrow ) experimental abilities / possibilities . however , this does nt mean we have to renounce to form a coherent picture of what s going on during the spin measurement , for instance by finding a meaningful structural analogy that could reveal us what could be the nature of the `` gaming apparatus '' in use . this is precisely what aerts did when he invented his numerous quantum - machines , which are ( idealized ) macroscopic ordinary objects able to reproduce , among other things , the puzzling non - kolmogorovian quantum probabilities . let us now describe the spin quantum - machine ( sqm ) and see how its functioning can simulate a spin-@xmath0 quantum measurement . the machine is very simple : it is constituted by a single point particle localized inside the shell of a three - dimensional ( euclidean ) empty sphere of unit radius , the possible states of which correspond to the different places it can occupy on its internal surface . let us assume that the point particle , in a given moment , is located in the position specified by the unit vector @xmath14 . we introduce a specific class of experiments @xmath38 , that can be performed on it , which are so defined . to carry out experiment @xmath38 , the following procedure has to be executed : a uniform sticky elastic band is stripped between the two opposite points @xmath20 and @xmath39 , therefore passing through the centre of the sphere [ fig . [ spin quantum machine ] , picture ( 1 ) ] . once the sticky elastic band is placed , one lets the point particle `` fall '' from its original location ( specified by @xmath14 ) orthogonally onto the elastic , and stick to it , in a given point [ fig . [ spin quantum machine ] , picture ( 2 ) ] . then , one waits until the elastic breaks , at some unpredictable point ( the rubber used to make the band is such that , once stretched , after a short time inevitably it breaks ) , and therefore the particle , which is attached to one of the two pieces of it , will be pulled to one of the opposite end points , @xmath20 or @xmath39 , thus producing the outcome of the experiment , i.e. , the final position - state of the particle , acquired as a result of the @xmath38-measurement [ fig . [ spin quantum machine ] , pictures ( 3 ) and ( 4 ) ] . of course , the experimenter can not know in advance in which point the elastic will break , as this depends on a number of fluctuating factors which are totally beyond her / his possibility of control . however , s / he knows that the elastic is uniform and therefore can reasonably assume that the probability that it will break in a given segment is proportional to the length of that segment , and under this ` natural ' hypothesis , s / he can easily calculate the probabilities associated to the two mutually exclusive outcomes . indeed , the probability that the particle ends up in point @xmath40 is given by the length @xmath41 of the piece of elastic between the particle and the end - point , divided by the total length of the elastic ( which is twice the unit radius ) . therefore , if @xmath42 is the angle between @xmath14 and @xmath20 , we obtain that the probability for outcome @xmath40 is given by : @xmath43 which exactly corresponds to the previously calculated quantum probabilities ( [ probability up2 ] ) and ( [ probability down ] ) , for measuring the spin @xmath44 of a spin-@xmath0 quantum entity prepared in state @xmath23 . so , the spin quantum machine , which is only made of ( idealized ) classical entities , is perfectly able to replicate the behavior of a spin-@xmath0 entity , i.e. , of a pure quantum entity , and produce the same probability calculus . the sqm model reveals a number of important features of quantum systems that we are now going to discuss , but before doing that , let us point out an important difference between a spin-@xmath0 quantum entity and the point particle of the sqm . when we want to determine the spin state of a quantum entity , all we can do is to perform measurements , for instance by means of a stern - gerlach apparatus whose magnetic field gradient is oriented along a given direction @xmath20 . in the same way , in the sqm , one disposes of experiments @xmath38 , which are the equivalent of the stern - gerlach measurements . but in the sqm one also has the possibility of looking in the machine and directly `` see '' the state position of the point particle , as one can usually do , in principle , with whatever classical corpuscle . this possibility , of directly looking in the machine and discover , at any time , what s happening , is precisely what confers the sqm its strong explicative power , i.e. , the possibility of fully visualizing the measurement process , as it evolves . however , such an `` insight '' should not be considered as a possibility to be used in practical terms , as an alternative to experiments @xmath38 , to determine the position - states of the point particle . as is the case with a spin-@xmath0 quantum entity , we must here do _ as if _ the only experiments at our disposal to `` find out '' what are the locations of the point particle on the sphere , are the @xmath38 ones , and no others . having said that , let us now observe what we can learn from the structural analogy offered by the sqm . _ observer effect_. a quantum measurement , in general , changes the state of the entity which is measured . this appears in very clear terms in the sqm : prior to the @xmath38-experiment the point particle is in a position - state @xmath14 , whereas at the end of it its state is either @xmath20 or @xmath39 . this means that a quantum measurement can not be considered as a process of mere _ discovery _ of properties that are already possessed by the system prior to the experiment , but as a process of _ creation _ of those same properties that are measured . to quote pascual jordan @xcite : `` observations not only disturb what has to be measured , they produce it [ ... ] . we compel [ the electron ] to assume a definite position [ ... ] . we ourselves produce the results of measurements . '' in other terms , the state ( [ spin state ] ) can not be interpreted as a description of the experimenter s lack of knowledge about whether the spin is `` up '' or `` down '' with respect to the @xmath3 direction , as these `` up '' and `` down '' outcomes are literally created during the interaction of the entity with the experimental apparatus , and were nt existing prior to it . of course , all this is already manifest in the quantum formalism , as is clear that states , represented by rays of the hilbert space , are typically changed by a measurement into other states , which are eigen - rays of the operator corresponding to the measurement . however , in the sqm we can explicitly observe that it is the _ invasiveness _ and _ unpredictability _ of the interaction between the entity and the measuring apparatus that is responsible for the creation of a condition that was nt actual before the experiment . _ non - commutativity_. the non - commutativity of certain quantum observables ( in the present case of @xmath45 and @xmath46 ) appears to be a direct consequence of the invasive character of quantum measurements . indeed , if what is measured is not discovered but created by the measurement , then the order with which one performs two subsequent measurements is going to deeply affect the final outcome , in the same way as putting on a sock , then a shoe , is not the same as putting on a shoe and then a sock . in the sqm , non - commutativity is evidenced by the fact that , if one performs an experiment @xmath38 ( the equivalent in the machine model of a measurement of spin component @xmath47 ) , followed by an experiment @xmath48 , the particle s final position will be either @xmath49 or @xmath50 , whereas if one performs first @xmath48 and then @xmath38 , the final position will be either @xmath20 or @xmath39 . therefore , if @xmath51 , these final states will necessarily be different , showing that the two experiments are incompatible , as their order of execution is relevant and can not be commuted . _ hidden measurements_. the nature of the interaction between the point particle and the elastic in the sqm model reveals a possible fascinating feature of quantum measurements . imagine for a moment that , instead of a uniformly breakable elastic band , one uses a band that can only break in a specific predetermined point @xmath52 ( the sphere s origin corresponding to @xmath53 ) . given the initial position - state @xmath14 of the point particle , it is clearly possible in this case to predict in advance if the outcome will be @xmath20 or @xmath39 . indeed , if point @xmath54 corresponds to position @xmath20 on the sphere , and @xmath55 to position @xmath39 , then for @xmath56 the outcome will certainly be @xmath20 , and @xmath39 for @xmath57 . an elastic band that breaks in a single predetermined point also constitutes an invasive experiment , altering the state of the point - particle , but it does so in a perfectly predictable way ( excluding of course the special case @xmath58 ) . let us call @xmath59 such a deterministic experiment . if we envisage an experiment whose procedure consists in choosing randomly one of the @xmath60 $ ] , then executing it ( which means that @xmath52 is considered to be a uniformly distributed random variable ) , then , again , we can only predict the outcomes in probabilistic terms , and the probabilities will be exactly given by ( [ probability up2 ] ) and ( [ probability down ] ) . this means that in the experiment @xmath38 ( i.e. , in the uniformly breaking elastic band ) a collection of deterministic ( potential ) experiments @xmath59 are in fact `` hidden , '' and the way in which one of these hidden deterministic experiments is selected during the measurement process is totally beyond the knowledge and possibility of control of the experimenter , as it depends on the presence of fluctuations in the experimental context . this situation has been called _ hidden measurements _ , by analogy with so - called _ hidden variables _ theories , where the lack of knowledge is hypothesized being in relation to the state of the system . now , in general terms , one can show that if a theory describes a situation where the lack of knowledge is about the state of the physical entity , then it is necessarily a classical statistical theory , obeying kolmogorov s axioms . on the other hand , if a theory describes a situation where the lack of knowledge is about the measurement to be actually performed , and such a measurement affects the state of the physical entity , then it is a non - classical statistical theory , violating classical kolmogorov s axioms , as quantum theory does @xcite . in other terms , what the spin quantum - machine strongly suggests , is that the fundamental difference between classical and quantum probabilities lies in the fact that the former describes a situation of lack of knowledge about the state of the system , whereas the latter describes a situations where there is knowledge of the state of the entity , but lack of knowledge about the interaction between the measurement apparatus and the entity . _ intermediate systems_. the spin quantum - machine is in fact a much richer model than what we have shown so far . indeed , it is very easy to modify its functioning to obtain probabilistic models that generalize the classical and quantum ones . to do this , aerts considered a generalized class of experiments , employing elastics of a more complex structure . let us call them @xmath61 , where @xmath62 is a real parameter comprised between @xmath63 and @xmath64 ( this modified spin quantum - machine model is then called @xmath62__-model _ _ @xcite ) . an @xmath61-measurement has the same protocol of a @xmath65-measurement , but this time the elastic band used is not anymore uniformly breakable , but breakable only in a segment of length @xmath66 around its center , and unbreakable in its lower and upper segments , as depicted in fig . [ spin epsilon machine ] ( we describe here a simplified version of the model , which was presented in @xcite ) . therefore , an @xmath67-experiment is a measurement using a fully uniformly breaking elastic band [ @xmath68 and corresponds to the pure quantum situation where there is _ maximum lack of knowledge _ about the point where the elastic is going to break . this is the situation described by the sqm , that we have previously analyzed , whose probabilities are given by ( [ probability up2 ] ) and ( [ probability down ] ) . on the other hand , an @xmath69-experiment is a measurement using an elastic band that breaks exactly in its middle , with certainty [ @xmath70 and corresponds to a purely classical ( deterministic ) situation with _ minimum lack of knowledge _ about the point where the elastic is going to break . but what about a general @xmath61-experiment , with @xmath71 , i.e. , an experiment using an elastic band which can only ( uniformly ) break around its center , in a segment of length @xmath66 ? the associated probabilities are easy to calculate , and one has to distinguish the following three cases : \(1 ) if the particle , when it `` falls '' orthogonally onto the elastic , lands on its upper unbreakable segment ( @xmath72 ) , then : @xmath73 \(2 ) if the particle , when it `` falls '' orthogonally onto the elastic , lands on its central uniformly breakable segment of length @xmath66 , ( @xmath74 ) , then : @xmath75 \(3 ) if the particle , when it `` falls '' orthogonally onto the elastic , lands on its lower unbreakable segment ( @xmath76 ) , then : @xmath77 to what kind of situation do the above probabilities correspond ? do they correspond to a classical or to a quantum probability model ? in fact , to none of them : they do nt fit into a classical kolmogorovian probability model , but neither into a non - kolmogorovian probability model of the quantum kind , as they truly correspond to new intermediate models , describing more general structures , of which the classical and quantum ones are limit cases @xcite . what is interesting here to emphasize , in the general analysis of the sqm model and of its @xmath62-model generalization , is that our knowledge about the behavior of the physical entity is clouded by the presence of fluctuations , whose dynamics is beyond our control power . in the ambit of the @xmath62-model the amplitude of these fluctuations can be quantified by the real parameter @xmath78 $ ] . when @xmath79 , the fluctuations which are responsible for the dispersion in the results of the measurement are maximal , as is maximal our lack of knowledge about the experimental outcome . this corresponds to the limit case of a pure quantum system . however , when @xmath62 decreases , but is not zero , we are in a situation of _ intermediate knowledge _ : for certain preparations of the physical system we can predict with certainty the outcome , whereas for others we are still in a situation where they can only be predicted in strict probabilistic terms . then , when @xmath62 reaches its lower value zero , we can say that the fluctuations in the measuring apparatus are in a sense smaller than the scale of the physical entity , and therefore can not affect it in terms of outcomes ( the different possible experiments that are randomly selected are not distinguishable in terms of their outcomes ) , so that again we are in a situation of maximal knowledge , typical of classical determinism . the above scheme describes a possible approach to the solution of the longstanding problem of finding a meaningful quantum - to - classical limit @xcite , as it provides the possibility of a continuous ( non - singular ) transition from these two limit situations . a last remark is due . usually it is believed that macroscopic entities , like the ordinary entities with which we interact daily , can not exhibit quantum ( or intermediate , quantum - like ) behaviors . as the sqm model clearly demonstrates , such a belief is unfounded . of course , for the reader it may be difficult to accept that a point particle on a sphere can also be considered a quantum entity . however , one should not forget that its quantum character is revealed when we perform on it certain experiments , and not others . we must not forget that the @xmath65 experiments are , in the sqm model , the only experiments that we are allowed , by definition , to execute . we are not allowed to `` see '' the point particle by other means , and this is the reason why , considering this restriction , the point particle can behave as a quantum entity . as aerts and durt emphasize @xcite : `` this is indeed exactly the situation that we encounter when we make investigations about quantum entities in the micro - world . we can not ` see ' or ` touch ' these entities , and have only knowledge about them by means of the experiments that we can carry out on them . '' before describing ( our alternative version , based on elastics , of ) aerts _ connected vessels of water _ model @xcite , let us briefly recall what bell s inequality is all about @xcite . bell was able to translate into a mathematical inequality certain general assumptions about physical systems , so that if these equalities are found to be experimentally violated , then at least one of the assumptions that are used to derive them must be wrong . we shall not prove here bell s inequality , but simply recall its expression . assume that , on a given physical entity , four different experiments can be performed . let us call them @xmath80 , @xmath81 , @xmath82 and @xmath83 . let us also call @xmath84 , @xmath85 , @xmath86 and @xmath87 the associated outcomes , which are assumed can only take the values @xmath88 or @xmath89 ( if the experiments are described by self - adjoint operators , as it is the case in quantum mechanics , then the outcomes correspond to their eigenvalues ) . a further hypothesis is that experiments @xmath80 and @xmath81 can also be performed together with either of experiments @xmath82 and @xmath83 , thus defining the additional _ coincidence _ experiments @xmath90 , @xmath91 , @xmath92 and @xmath93 . to each coincidence experiment @xmath94 , @xmath95 , @xmath96 , one can associate the expectation value @xmath97 of the product of outcomes @xmath98 , by : @xmath99 where @xmath100 is the probability that the coincidence experiment @xmath94 yields the outcomes @xmath101 . then , under the assumption that the outcomes are independently determined by some hidden variables , so that the expectation ( [ expectation value ] ) can be written as the integral of the product of the two outcomes over these hidden variables ( an assumption often referred to as _ bell locality _ ) , it is possible to prove the following ( bell ) inequality @xcite : latexmath:[\[\label{bell inequalities } bell derived this inequality having in mind the physical system originally introduced by bohm @xcite , of two entangled spin-@xmath0 quantum entities in a _ singlet ( zero ) spin state _ , which is an _ entangled _ ( non product ) state : @xmath103 and @xmath104 describe `` up '' and `` down '' eigenstates of the spin operator @xmath105 , along the @xmath20-direction , in two separated ( and arbitrarily distant ) regions of space @xmath106 and @xmath107 , respectively . on this singlet spin state , a certain number of spin measurements can be performed , using for instance two adjustable stern - gerlach filters , and the associated screen detectors , placed in regions @xmath106 and @xmath107 , as schematically depicted in fig . [ coincidence ] . in this ambit , one can conceive the following four experiments : @xmath108 , @xmath109 , @xmath110 and @xmath111 , where @xmath112 , @xmath113 , @xmath114 and @xmath115 are unit vectors describing the orientations of the stern - gerlach filters ( orthogonal to the direction of flight of the pair of quantum entities forming the singlet state ) , @xmath116 is the identity operator , @xmath117 is the tensor product , and the symbol @xmath118 is to be interpreted as `` the experiment on the left hand side corresponds to the measuring of the observable on the right hand side . '' these experiments ( that can take values @xmath88 or @xmath89 ) can be combined to define the coincidence experiments : @xmath119 where @xmath120 , @xmath121 . now , considering the quantum mechanical expectation value @xmath122 , and using the properties of the spin operators ( see any good book of quantum mechanics ) , one can show that @xmath123 . therefore , if one takes @xmath112 , @xmath113 , @xmath114 and @xmath115 to be coplanar and choose the angles between @xmath112 and @xmath114 , @xmath114 and @xmath113 , and between @xmath113 and @xmath115 to be equal to @xmath124 , one obtains : @xmath125 which is clearly a violation of ( [ bell inequalities ] ) . this result shows that no physical theory which is local in the sense specified by bell , can agree with all the statistical implications of quantum mechanics . bell locality hypothesis can be understood as an assumption about the _ experimental separation _ of the two ( here spin ) entities emerging from the source and which , after an arbitrary long time of flight , are detectable in two regions @xmath106 and @xmath107 that are _ spatially separated _ by an arbitrarily large distance . in accordance with einstein s view , the assumption of many physicists was that a spatial separation would also imply an experimental separation , so that the expectations were that the microphysical reality would obey bell s inequality . but these expectations were disregarded , as we today all know , by the historical experiments with entangled pairs of aspect et al . @xcite ( who considered the equivalent situation of photons polarization measurements , instead of spin measurements ) , and the many others that since then followed , who confirmed the reality of the violation ( [ bell violation quantum ] ) , and therefore the correctness of the quantum formalism describing entangled states . in other terms , we have to renounce to local realism `` la bell '' in the description of the microworld . but , how can we _ understand _ then the origin of the violation of bell s inequality and , can they be violated only by microscopic quantum entities ? as we shall see in the next section , bell s inequality is easily violated also by macroscopic ` classical ' entities , provided we experiment on them in a very specific way . this will offer us the opportunity to see what exactly happens when a definite value of a property of the system in region @xmath106 is acquired by virtue of a measurement carried out in region @xmath107 , and vice versa . let us now describe a macroscopic entity that can maximally violate bell s inequality ( [ bell inequalities ] ) . as we have mentioned in the introduction , aerts considered a system made of vessels of water connected by tubes and siphons @xcite . here we shall present a much simpler system , which can play the same role as the connected vessels : a single uniform elastic band . we assume that the uniform elastic band has been made with a _ red _ rubber and that when unstreched , its length is @xmath126 . two scientists are placed at the two ends of it ( let us call them @xmath106 and @xmath107 , respectively ) , and can perform certain experiments . scientist @xmath106 can perform on the left end of the elastic band two experiments , @xmath80 and @xmath81 , which are defined as follow . experiment @xmath80 consists in grabbing the end of the elastic band and pulling it with force , then measuring the length of the unstreched elastic band that has been collected in this way : if it is greater than @xmath127 , the outcome @xmath128 , otherwise @xmath129 . experiment @xmath81 is much simpler as it consists in simply looking at the elastic band and checking whether its color is red . if it is so , then the outcome @xmath130 , otherwise @xmath131 . scientist @xmath107 can perform the same experiments as scientist @xmath106 , but on the right end of the uniform elastic band . in other terms , @xmath82 is defined as @xmath132 and @xmath83 as @xmath133 ( see the schematic representation of fig . [ coincidence with elastic ] ) . of course , since the elastic band is made with a red rubber and its length is @xmath126 , all of the four above mentioned experiments , when performed individually , can only produce the outcome @xmath88 ( as is clear that when only one experimenter is pulling the elastic , the whole of it will come in her / his hands ) . the same remains obviously true when the coincidence experiments @xmath91 , @xmath92 and @xmath93 are performed at the same time by the two scientists , whose outcomes are always @xmath134 ( we observe that when the elastic is pulled by one of the experimenters , the other one still has the possibility to locally check its color , as is clear that the light coming from the elastic is always available to be detected at her / his location ) . the situation changes however when one considers the coincidence experiment @xmath90 , which creates correlations . indeed , if the two scientists grab and pull the elastic band at the same time , it will break in some unpredictable point , so that if one scientist collects a fragment of length greater than @xmath127 , the other one will necessarily collect a fragment of length lower than @xmath127 . thus , the only possible outcomes of the coincidence experiment @xmath90 are @xmath135 and @xmath136 , and of course they have the same probability to occur , which is equal to @xmath0 . according to ( [ expectation value ] ) , we therefore obtain that @xmath137 , and @xmath138 , so that : @xmath139 in other terms , not only the elastic band can be used to break bell s inequality , but it can do it much more than the typical example of two coupled spin-@xmath0 entities in a singlet state . in @xcite david mermin mentions a conversation with a distinguished physicist in princeton . he asked him how he thought einstein would have reacted to the violation of bell s inequalities : `` he said that einstein would have gone home and thought about it hard for several weeks that he could nt guess what he would then have said , except that it would have been extremely interesting . '' this author likes to think that einstein might have tried to find out an example `` la aerts , '' like his _ connected vessels model _ , or the equivalent _ elastic band model _ that we have described . and based on an example of this kind , he would have probably reached the same conclusions as aerts did , regarding the real mystery hidden in bell s inequalities ( a.k.a . bell s no - go theorem ) . what s immediately evident in the elastic band model is the fact that the entity playing the role of the quantum singlet state the uniform elastic band is a single genuine whole entity , and not two entities , and certainly can not be understood as the sum of two separated interacting parts , each one described by its own specific state . indeed , the elastic band is present not only in the two regions @xmath106 and @xmath107 , where the two scientists are located ( close to its ends ) , but also in between these two regions : it possesses what has been called the property of _ macroscopic wholeness _ @xcite , i.e. , the property of _ hanging together through space _ , which means it can not be localized in different macroscopically separated spatial regions without also being localized somewhere in the region ` in between ' them . this wholeness aspect was already pointed out by schroedinger @xcite , when he emphasized that entangled states like ( [ singlet state ] ) are associated to a notion of _ non - separability _ , in the sense that two quantum entities can find each other in a state such that only the properties of the pair appear to be defined , whereas the individual properties of each one of the two entities that have formed the pair remain totally undefined . clearly , the notion of non - separability introduced by schroedinger becomes fully explicit and viewable with the elastic band example . indeed , one can certainly confer well defined properties to the two distant ends of the unbroken elastic band , like their spatial distance , but on the other hand we can not attribute to each of these distant ends , considered `` separately , '' well - defined properties , as they do nt possess any independent reality outside of the reality of the whole band . but when the coincidence experiment @xmath90 is executed , something very special happens : two entities are _ created _ and correlations that were nt present before the experiment are in this way actualized . this creation of correlations , that can violate bell s locality hypothesis ( and therefore bell s inequality ) , can not be described in terms of local hidden variables , associated to the individual states of the ( only potentially existing ) separated fragments of the elastic band , as the correlations are produced by the measurement process itself , which in the present case is a _ hidden measurement _ process . as we observed for the sqm model , a hidden measurement process is a process characterized by two ingredients : ( 1 ) an aspect of creation of new properties and ( 2 ) an aspect of indeterminism , i.e. , the fact that the outcome of such a creation process can not be predicted in advance . therefore , a natural question arises : is it the creation of correlations that were inexistent before the experiment that is responsible for the violation of bell s inequality , or is it the indeterminism that is inherent in this process , or is it both ? this question was deeply analyzed in @xcite , using a much more complex macroscopic model than the elastic band model presented here , able to violate bell s inequality not in a maximal way , but exactly in the same way a quantum singlet state does @xcite . by varying in the model two parameters , quantifying respectively the degree of correlation and of indeterminism present in the system , the authors were able to show that the crucial ingredient in the violation of bell s inequality is not the indeterminism , but the correlation , i.e. , that it is the non - local aspect expressed by the correlation that is the true source of the violation , and that the presence of indeterminism can in fact decrease the value bell s inequality takes . without engaging in the rather involved analysis of the more complex situation presented in @xcite , let us however show , in our simpler elastic band model , how bell s inequality can be violated even when all source of indeterminism in the experiments has been eliminated . to do this , we simply have to replace the hidden measurement promoted by the uniform elastic band , by a `` pure '' ( non - hidden ) measurement promoted by a ( red ) elastic band that can only break ( if the elastic is stretched with sufficient force ) in a predetermined point , located , say , at distance @xmath140 from its left end . then , the outcomes of the coincidence experiments @xmath91 , @xmath92 and @xmath93 will be again @xmath134 , as it was the case for the uniform elastic band ( scientists can not break the elastic if they pull it only from one end ) . on the other hand , concerning experiment @xmath90 , the only difference is that now we can predict in advance what will be the outcome , meaning that the correlation is created in a perfectly deterministic way , always producing the outcome @xmath136 . hence , as for the uniform elastic band , we find that @xmath137 , and @xmath138 , so that ( [ bell violation classical ] ) holds and bell s inequality is again maximally violated . this shows that it is the _ creation of correlations _ which is really responsible for the violation , and not its deterministic or indeterministic character . correlations that were not present before the experiment , but are created by and during it , are called _ correlations of the second kind _ @xcite . this to distinguish them from so - called _ correlations of the first kind _ , which are already present before the experiment is executed , and are therefore not created but only discovered by it . let us observe that although in the elastic experiment the correlation ( of the second kind ) is created instantly , in the precise moment when elastic breaks , the information relative to the correlation can not travel faster than the speed of contraction of the elastic . in other terms , the elastic ca nt be used to produce superluminal signaling and einstein locality is obviously preserved . let us also observe that , as explained in @xcite , one can not use correlations of the first kind to violate bell s inequality . let us show this explicitly in our model by replacing the whole elastic entity by an already broken elastic entity , where the left fragment is , say , of length @xmath140 and the right fragment of length @xmath141 ( see fig . [ coincidence with broken elastic ] ) . clearly , the respective lengths of the two elastic fragments are correlated , as is clear that their sum must be equal to the total length @xmath126 of the ( previously ) unbroken elastic band . however , this correlation is now of the first kind , as it is already present in the system before the experiments are carried out . also in this case , the outcomes of all the experiments are predetermined , as it was the case for the ( unbroken ) elastic band only breakable in one point , and we have that the outcomes of @xmath92 and @xmath93 are @xmath134 , and the outcome of @xmath90 is @xmath136 . on the other hand , this time experiment @xmath91 has outcome @xmath136 , so that @xmath142 , and @xmath143 , yielding : @xmath144 in plain accordance with bell s inequality ( [ bell inequalities ] ) . we can therefore conclude with aerts that @xcite : `` the violation of bell inequalities is not a property of micro - entities . bell inequalities can equally well be violated by coincidence experiments on classical microscopic entity . in fact bell inequalities can always be violated if during the coincidence experiments one breaks one entity into separated pieces , and by this act creates the correlations . '' we call attention to the fact that , because of its connectedness , a whole elastic band is the expression of a number of relations between its different parts . each point of the elastic is indeed representative of a specific relation between the two non - separated parts of the elastic which are on the left and right sides of the point ( the relation being expressed by the fact that the sum of the unstreched lengths of these two parts has to be @xmath126 ) . hence , one may be tempted to conclude that the violation of bell s inequality is in fact the result of correlations of the first kind , between _ non - separated parts _ , and not of correlations of the second kind , between _ separated fragments_. this however would be a wrong conclusion , as only correlations between separated fragments are observed during the experiments , and these correlations are literally created by them . indeed , the non - separated parts of an unbroken elastic band only correspond to _ potential separated fragments _ , and therefore to _ potential correlations _ , when considered in relation to the outcomes of the experiments . this remark is important in order to understand that entangled states , like singlet states , are the expression of a ( hidden ) connected structure of relations out of which correlations between outcomes are created , not discovered . this is so because the properties which are observed during the coincidence experiments do not exist prior to them , but are actualized by them , in the same way as separated fragments do not exist until the whole elastic is broken . this mechanism of creation of correlations is already manifest in the quantum formalism , as is clear that the singlet state is a state of zero spin , which is rotational invariant , so that there is nothing special about the direction @xmath20 we have chosen to represent the state in terms of `` up '' and `` down '' components : if one chooses another arbitrary direction , say @xmath49 , then by a direct algebraic calculation one can show that ( [ singlet state ] ) can be equivalently written as : latexmath:[\[\label{singlet state bis } this means that the correlation between the outcomes of the left ( region @xmath106 ) and right ( region @xmath107 ) spin - measurements does nt depend on the global orientation chosen for the two spin filters ( only their relative orientation is important ) , and this also means that @xmath146 does nt describe the state of two - entities having already actualized their spin , although the way it is mathematically written may wrongly suggest so . in fact , their spins are literally _ created _ by the coincidence experiment , which in a sense `` breaks '' the singlet state into a state describing two separated spin entities with a specific direction ( although now entangled with the experimental apparatus ) , exactly in the same way an elastic band of length @xmath126 does nt describe two actually separated elastic bands , but only two potentially separated elastic bands , which can be created ( in a deterministic or indeterministic way ) by performing a correlation experiment of the breaking kind . let us briefly summarize the main results we have presented in this paper . to understand what could possibly happen `` behind the scenes , '' during a quantum measurement , we have reviewed two of aerts macroscopic models : the sqm ( and its @xmath62-model generalization ) @xcite , and an elastic band model inspired by a previous connected vessels of water model @xcite . in this way , we have gained considerable insight into the functioning of the quantum level of our reality , which therefore appears to be less mysterious than expected , considering that also ordinary macroscopic entities can give rise to quantum probabilities , or violate bell s inequalities , and that the way they do it is perfectly under our eyes . more precisely , thanks to the sqm model , and the structural analogies it provides , we have explained the emergence of quantum probabilities in terms of hidden , invasive measurements , which are selected in a way that can not be controlled and therefore predicted by the experimenter . this of course does nt mean that we now understand everything about quantum measurements , which surely retain part of their mystery . indeed , the sqm model just tells us in what direction one should search , but not what exactly to find : we have to search not for hidden variables associated to the state of the entity , as it has been historically done , but for hidden variables associated to the measurement process , i.e. , for the `` pure '' ( possibly deterministic ) measurement interactions which are selected through a ( symmetry breaking ) mechanism that can not be controlled ( and therefore be known ) by the experimenter . with regard to the nature of these hidden measurements , as far as the present author can judge , the mystery remains intact . of course , there exist unconventional approaches to quantum mechanics where some attempts have been made to identify possible ( real ) processes that would operate at a subquantum level , and that could possibly explain the emergence of quantum probabilities . since we have quoted john cramer in the introduction , let us mention , as a simple paradigmatic example , his _ transactional interpretation _ ( ti ) @xcite , which has been recently revisited by ruth kastner @xcite . ti is a time - symmetric re - interpretation of conventional quantum mechanics , such that kets are associated to retarded `` offer waves of possibilities , '' evolving forward in time , emitted by entities which , in a given context , play the role of sources , whereas bras are associated to advanced `` confirmation waves of possibilities , '' evolving backward in time , emitted by entities playing the role of absorbers , in response to the offers received from the sources . these a - temporal offer - confirmation exchanges between sources and potential absorbers , give rise to so - called _ incipient transactions _ , which are then selected ( through a symmetry breaking mechanism ) to give rise to _ actualized transactions _ , i.e. , to processes that select specific outcomes in the form of actual transfers of given quanta of energy and momentum . it is not our intention to enter here into the details of the ti and discuss the strengths and weaknesses of this rather unusual interpretation . our only intention was to provide a very simple example of an approach whose ontology contemplates a candidate for something like a hidden measurement interaction , which in the ti case would correspond to the process of actualization of incipient transactions . regarding the hidden - measurement approach , some readers may be tempted to believe that , because of the well - known gleason s theorem @xcite and kochen - specker s impossibility proof @xcite , it would be unfeasible to construct models like the sqm , for hilbert spaces of dimension greater than 2 . this is however not the case , as no - go theorems for hidden variables only apply to models with hidden variables referring to the _ state of the system _ , and not to models where the hidden variables refer instead to the _ measurement process _ ( see the discussion in @xcite and the references cited therein ) . in other terms , although one could consider , in light of the above results , that the sqm model would not per se be sufficient to possibly explain the origin of quantum probabilities , one should bear in mind that its built - in hidden - measurement mechanism can be easily generalized to describe the probability model of quantum systems of arbitrary dimension , as it was done by aerts in @xcite . a discussion of these more general situations would however go beyond the scope of the present paper , and we refer the interested readers to @xcite , and the references therein . let us now comment on the second part of the present paper : the violation of bell s inequalities . thanks to the elastic band model , we can easily affirm that much of the strangeness associated to these phenomena , which disturbed so much einstein , schroedinger , bell and many others , is gone . indeed , it is possible to explain quantum correlations , the `` spooky actions at a distance , '' as einstein used to call them , as processes during which a whole entity is broken into parts , in a deterministic or indeterministic way . again , this does nt mean we now understand everything about the quantum correlations produced by entangled states . indeed , as we explained , for a macroscopic entity to be able to violate bell s inequality , the _ conditio sine qua non _ is that it possesses the property of _ macroscopic wholeness _ : it has to be present in the spatially separated regions @xmath106 and @xmath107 , where the coincidence experiments are performed , but it has also to be present in the region between them . but quantum microscopic entities do not possess this macroscopic wholeness property , and this certainly constitutes that part of the mystery that remains intact when considering the violation of bell s inequality at the microscopic level , in epr - like experiments . indeed , considering a singlet state formed from a pair of electrons that have flown apart , and have reached the distant regions @xmath106 and @xmath107 , respectively , we know that one electron can be detected with probability close to one in region @xmath106 , and same for the other electron in region @xmath107 , but that the probability of detecting them in between is almost zero . so , it seems that at the microscopic level , spatially separated entities can nevertheless remain interconnected , not through space , but in some other , yet to be explained , way . of course , the difficulty many physicists experience in truly understanding this possibility lies once more in the difficulty one might have in visualizing a concrete model that would render it manifest . this however is not unfeasible . a possibility has been proposed in @xcite , where a generalization of the concept of macroscopic wholeness was proposed , called _ process - macroscopic wholeness _ , expressing the idea that two entities can remain connected not only through space , in a static way , but also _ through time _ , in a more dynamic way . these kind of explanations force us to abandon an old preconception , which may be the one truly hindering our understanding of the microworld : that microscopic entities should always be present in our three - dimensional space and that , more generally , reality should only exist within space . on the contrary , it seems reasonable to hypothesize that microscopic entities are non - spatial entities , and that , quoting aerts , space would only be @xcite `` [ ... ] a momentaneous crystallization of a theatre for reality where the motions and interactions of the macroscopic material and energetic entities take place . but other entities - like quantum entities for example - ` take place ' outside space , or - and this would be another way of saying the same thing - within a space that is not the three dimensional euclidean space . '' clearly , we can not enter here in the analysis of these important ideas ( for more details , see for instance @xcite and the references cited therein ) , as this would bring us too far away from the primary motivation of this paper , which was only to conceptually review two of aerts historical machine - models , in order to promote a deeper understanding of the physical content `` hidden '' within the abstract quantum formalism . it is worth emphasizing that the truly interesting aspect about these models is not their ability to realistically describe physical entities as such , but to capture , by means of powerful structural analogies , the possible logic at the basis of their functioning , when they interact with the different experimental contexts . also , these machine - models force us to abandon too simplistic distinctions about classical and quantum entities . indeed , the quantum behavior of a physical entity appears to also depend on how we choose to actively experimenting on it , according to specific protocols . and this is the reason why also macroscopic ordinary entities can behave in a quantum - like fashion . this remark has some relevance also in relation to possible _ loopholes _ in experiments testing bell s inequality , where by loophole we mean here a more or less consciously used assumption which would undermine the validity of an experiment ( see for instance @xcite , pages 600 - 605 ) . the analysis here presented highlights one of these possible loophole : the ( wrong ) assumption that bell s inequality could only be violated by microscopic systems and that the violation would be an indication of the presence of only microscopic quantum properties . on the contrary , we know that , quoting once more aerts , @xcite `` bell inequalities can be genuinely violated in situations that do not pertain to the microworld . of course , this does not decrease the peculiarity of the quantum mechanical violation in the eprb experiment . what it does , is shed light on the possible underlying mechanisms and provide evidence that the phenomenon is much more general than has been assumed . '' in other words , also the detecting apparatus , in a typical coincidence experiment , when considered as an interconnected macroscopic whole entity , is in principle able to produce correlations of the second kind , which could be in principle at the origin of the observed violation of bell s inequality . this possibility is sometimes called the `` locality loophole . '' @xcite to close this loophole , one has to design the coincidence experiments in such a way to be sure that only the correlations originating from the entangled microscopic entity are at the origin of the violation of bell s inequality . for this , one has to exploit the fact that , because of the non - spatial nature of microscopic entities , the effects resulting from the creation of correlations by entangled states `` travel '' virtually instantaneously ( certainly thousands of times faster than the light speed @xcite , although this can not be used to produce superluminal signaling @xcite ) , whereas possible effects due to correlations resulting from the interconnectedness of the detectors certainly can not be communicated at speeds greater than the speed of light . this was done by aspect , in his famous polarization experiments with entangled photons ( see @xcite and the references therein ) , where the choice of orientations of the polarizers were changed randomly in very short times , during the flight of the two entangled photons , in order to dramatically enforce relativistic separation . thanks to these experiments , it was possible to conclude , quoting aspect , @xcite `` [ ... ] that an entangled epr photon pair is a non - separable object ; that is , it is impossible to assign individual local properties ( local physical reality ) to each photon . in some sense , both photons keep in contact through space and time . '' to conclude , we point out that many other machine models have been invented in addition to those mentioned in this article , and we can only invite the reader to take cognizance of them , to get advantage of the numerous insights they are able to promote . let us mention aerts model describing subsystems connected by rigid rods , which can violate bell s inequalities in a non - maximal way , as microscopic quantum entities can do @xcite , also revealing possible unexpected interpretations of mixed states in the standard quantum formalism @xcite . and let us also mention a recent model that has been presented by this author , called the @xmath147-quantum machine , which is able to replicate the scattering probabilities of a one - dimensional quantum scattering process by a dirac delta - function potential . but aerts , in his attempt to truly understand quantum mechanics , has not only invented mechanistic models . in more recent years he has also devoted a considerable effort in using the quantum mechanical formalism to model , with great success , human conceptual situations as they appear in cognition , decision theory and economics . this led him to ask a deep and thought provoking question @xcite : `` if quantum mechanics as a formalism models human concepts so well , perhaps this indicates that quantum particles themselves are conceptual entities ? '' this question was the starting point of a new interpretation of quantum mechanics , which is probably today s most advanced explanatory framework to understand this puzzling theory . according to aerts , quantum entities would @xcite `` [ ... ] interact with ordinary matter , nuclei , atoms , molecules , macroscopic material entities , measuring apparatus , ... , in a similar way to how human concepts interact with memory structures , human minds or artificial memories . '' therefore , quoting again aerts @xcite : `` if proven correct , this new quantum interpretation would provide an explanation according to which ` quantum particles ' behave like something we are all very familiar with , and have direct experience with , namely concepts . '' we will certainly not comment any further this subtle explanatory framework and its effectiveness in truly explaining phenomena such as entanglement and non - locality , which are traditionally considered ( in the spirit of feynman s quote cited in the introduction ) as `` not understood , '' and leave to the reader the intellectual pleasure to directly discover them in aerts recently published articles @xcite . let us just observe that it is rather remarkable that what is probably the best available model to understand the mysteries of the quantum level of our reality has always been at hand , hidden in the very structure of our cognitive processes . i m pleased to dedicate this article to diederik aerts , a scientist of our times who , like the giants of the past , is promoting through his multi - faceted research the possibility of reaching a true understanding of the working of our mysterious reality . 41 j. g. cramer , the transactional interpretation of quantum mechanics , rev . phys . , 58 , 647 ( 1986 ) . r. p. feynman , the character of physical law , penguin books ( 1992 ) . w. heisenberg , philosophic problems of nuclear science , trans . f. c. hayes ( 1952 ) , 38 . m. sassoli de bianchi , ephemeral properties and the illusion of microscopic particles , found . science , 16 , 4 , 393409 ( 2011 ) . m. sassoli de bianchi , from permanence to total availability : a quantum conceptual upgrade , to appear in : foundations of science ; doi : 10.1007/s10699 - 011 - 9233-z . d. aerts , the stuff the world is made of : physics and reality , 129 , in : the white book of ` einstein meets magritte ' , edited by : diederik aerts , jan broekaert and ernest mathijs , kluwer academic publishers , dordrecht ( 1999 ) . d. aerts , quantum mechanics : structures , axioms and paradoxes , 141 , in : the indigo book of ` einstein meets magritte ' , edited by : diederik aerts , jan broekaert and ernest mathijs , kluwer academic publishers , dordrecht ( 1999 ) . d. aerts , a possible explanation for the probabilities of quantum mechanics , j. math , phys . , 27 , 202210 ( 1986 ) . d. aerts , quantum structures : an attempt to explain the origin of their appearance in nature , int . phys . , 34 , 1165 ( 1995 ) . d. aerts , the entity and modern physics : the creation - discovery view of reality , in : interpreting bodies : classical and quantum objects in modern physics , edited by : e. castellani , princeton unversity press , princeton ( 1998 ) . d. aerts , quantum structures due to fluctuations of the measurement situations , int . j. theor . phys . , 32 , 22072220 ( 1993 ) . a. d. oconnell et al , quantum ground state and single - phonon control of a mechanical resonator , nature 464 , pp . 697 - 703 ( 2010 ) . d. aerts , an attempt to imagine parts of the reality of the micro - world , 325 , in : problems in quantum physics ii ; gdansk 89 , edited by : j. mizerski et al . , world scientific publishing company , singapore ( 1990 ) . d. aerts , the missing element of reality in the description of quantum mechanics of the epr paradox situation , helv . acta , 57 , 421428 ( 1984 ) . d. aerts , a possible explanation for the probabilities of quantum mechanics and example of a macroscopic system that violates bell inequalities , in : recent developments in quantum logic , edited by : p. mittelstaedt and e. w. stachow , grundlagen der exacten naturwissenschaften , band 6 , wissenschaftverlag , bibliografisches institut , mannheim ( 1985 ) . quoted by m. jammer , the philosophy of quantum mechanics , wiley , new york ( 1974 ) p.151 . d. aerts , the origin of the non - classical character of the quantum probability model , in : information , complexity , and control in quantum physics , edited by : a. blanquiere et al , springer - verlag ( 1987 ) . d. aerts and t. durt , quantum , classical and intermediate , an illustrative example , found . phys . , 24 , p. 1353 d. aerts and t. durt , quantum , classical and intermediate : a measurement model , in : the proceedings of the international symposium on the foundations of modern physics , helsinki , finland , edted by : c. montonen , et al . , editions frontieres , gives sur yvettes , france ( 1994 ) . a.m. gleason , measures on the closed subspaces of a hilbert space , j. math . 6 , 885893 ( 1957 ) . s. kochen and e.p . specker , the problem of hidden variables in quantum mechanics , j. math . mech . , 17 , 5987 ( 1967 ) . d. aerts , b. coecke , b. dhooghe and f. valckenborgh , a mechanistic macroscopic physical entity with a three - dimensional hilbert space description , helv . acta , 70 , 793802 ( 1997 ) . d. aerts , s. aerts , j. broekaert and l. gabora , the violation of bell inequalities in the macroworld , found . phys . , 30 , pp . 1387 - 1414 ( 2000 ) . d. bohm , quantum theory , prentice - hall , englewood cliffs , new york ( 1951 ) . j.s . bell , in : foundations of quantum mechanics , p. 171 , proceedings of the international school of physics `` enrico fermi , '' course xlix , edited by : b. despagnat , academic press , new york ( 1971 ) . j. s. bell , on the einstein - podolsky - rosen paradox , physics ( long island city , n.y . ) , 1 , p. 195 ( 1964 ) . reproduced as ch . 2 of j. s. bell , speakable and unspeakable in quantum mechanics ( cambridge university press , 1987 ) . a. aspect et al . , experimental realization of einstein - podolsky - rosen - bohm gedankenexperiment : a new violation of bell s inequalities , phys . , 49 , p. 91 a. aspect , bell s inequality test : more ideal than ever , nature ( london ) , 398 , p. 189 n. d. mermin , is the moon there when nobody looks ? reality and the quantum theory , physics today , pp . 38 - 47 , april ( 1985 ) . s. aerts , a realistic device that simulates the non - local pr box without communication , arxiv : quant - ph/0504171 ( 2005 ) . e. schroedinger , die gegenwartige situation in der quantenmechanik , naturwissenschaftern , 23 , pp . 807 - 812 ( 1935 ) . english translation : john d. trimmer , proceedings of the american philosophical society , 124 , 323 - 38 ( 1980 ) . reprinted in : quantum theory and measurement , edited by : j. a. wheeler and w. h. zurek , princeton university press , princeton , p. 152 d. aerts , a mechanistic classical laboratory situation violating the bell inequalities with @xmath148 , exactly ` in the same way ' as its violations by the epr experiments , helv . acta , 64 , 1 - 23 ( 1991 ) . r. e. kastner , the quantum liar experiment in cramer s transactional interpretation , studies in history and philosophy of modern physics , 41 ( 2 ) pp . 86 - 92 , may ( 2010 ) . d. aerts , a mechanistic classical laboratory situation violating the bell inequalities with @xmath148 , exactly ` in the same way ' as its violations by the epr experiments , helv . acta , 64 , 124 ( 1991 ) . aerts , d. , the description of joint quantum entities and the formulation of a paradox , int . j. theor . , 39 , pp . 485496 ( 2000 ) . m. sassoli de bianchi , the @xmath147-quantum machine , the @xmath149-model , and the non - ordinary spatiality of quantum entities , to appear in : foundations of science , arxiv:1104.4738 [ quant - ph ] . sassoli de bianchi , m. , `` ephemeral properties and the illusion of microscopic particles , '' foundations of science , 16 , no . 4 pp . 393409 ( 2011 ) ; doi : 10.1007/s10699 - 011 - 9227-x . m. sassoli de bianchi , `` from permanence to total availability : a quantum conceptual upgrade , '' to appear in : foundations of science , doi : 10.1007/s10699 - 011 - 9233-z . d. aerts , quantum particles as conceptual entities . a possible explanatory framework for quantum theory , foundations of science , 14 , pp . 361411 ( 2009 ) . d. aerts , interpreting quantum particles as conceptual entities , int . phys . , 49 , pp . 29502970 ( 2010 ) . g. auletta , m. fortunato , and g. parisi , _ quantum mechanics _ , cambridge university press ( 2009 ) . kolmogorov , foundations of the theory of probability . translation edited by n. morrison . chelsea publishing company , new york ( 1956 ) . d. salart et . al . , testing the speed of spooky action at a distance, nature 454 ( 7206 ) : 861 - 864 ( 2008 ) . h. p. hberhard , bell s theorem and the different concepts of locality , nuovo cimento , 46b , pp . 392419 ( 1978 ) .	from the beginning of his research , the belgian physicist diederik aerts has shown great creativity in inventing a number of concrete machine - models that have played an important role in the development of general mathematical and conceptual formalisms for the description of the physical reality . these models can also be used to demystify much of the strangeness in the behavior of quantum entities , by allowing to have a peek at what s going on in structural terms behind the `` quantum scenes , '' during a measurement . in this author s view , the importance of these machine - models , and of the approaches they have originated , have been so far seriously underappreciated by the physics community , despite their success in clarifying many challenges of quantum physics . to fill this gap , and encourage a greater number of researchers to take cognizance of the important work of so - called _ geneva - brussels school _ , we describe and analyze in this paper two of aerts historical machine - models , whose operations are based on simple breakable elastic bands . the first one , called the _ spin quantum - machine _ , is able to replicate the quantum probabilities associated with the spin measurement of a spin-@xmath0 entity . the second one , called the _ connected vessels of water model _ ( of which we shall present here an alternative version based on elastics ) is able to violate bell s inequality , as coincidence measurements on entangled states can do .
, for the two regions of --space indicated in the figure ( a ) , and @xmath0 for @xmath1 compared to the predictions of the lo mc generators phojet and pythia ( b ) . figures ( c ) and ( d ) show the value of @xmath2 as a function of the transverse energy and pseudo - rapidity of the jet respectively , compared to the pythia prediction . , scaledwidth=47.0% ] at @xmath3 as a function of the jet transverse energy for @xmath4 ( a ) and @xmath5 ( b ) , and as a function of the jet pseudo - rapidity for @xmath4 ( c ) and @xmath5 ( d ) . in each figure the results obtained using the inclusive and the cone jet algorithm are shown and compared to the pythia prediction.,scaledwidth=47.0% ] we have studied the production of di - jets in the collisions of two quasi - real photons at an centre - of - mass energy from 189 to 209 gev , with a total integrated luminosity of collected by the opal detector at lep . di - jet events are of particular interest , as the two jets can be used to estimate the fraction of the photon momentum participating in the hard interaction , which is a sensitive probe of the structure of the photon . the transverse energy of the jets provides a hard scale that allows such processes to be calculated in perturbative qcd . fixed order calculations at next - to - leading order ( nlo ) in the strong coupling constant @xmath6 for di - jet production are available and are compared to the data , providing tests of the theory . leading order monte carlo ( mc ) generators are used to estimate the importance of soft processes not included in the nlo calculation . more details on the results presented here can be found in @xcite . the -clustering algorithm @xcite is used as opposed to the cone algorithm @xcite in our previous publications @xcite for the measurement of the differential cross - sections , because of the advantages of this algorithm in comparing to theoretical calculations @xcite . the cone jet algorithm is used to demonstrate the different structure of the cone jets compared to jets defined by the -clustering algorithm . at colliders the photons are emitted by the beam electrons ( positrons ) . most of these photons carry only a small negative four - momentum squared , @xmath7 , and can be considered quasi - real ( @xmath8 ) . the electrons are hence scattered with very small polar angles and are not detected . events where one or both scattered electrons are detected are not considered in the present analysis . in lo qcd , neglecting multiple parton interactions , two hard parton jets are produced in interactions . in single- or double - resolved interactions , these jets are expected to be accompanied by one or two remnant jets . a pair of variables , and , can be defined @xcite that estimate the fraction of the photon s momentum participating in the hard scattering : @xmath9 where @xmath10 is the momentum component along the @xmath11 axis of the detector and @xmath12 is the energy of the jets or objects of the hadronic final state ( hfs ) . in lo , for direct events , all energy of the event is contained in two jets , i.e. , @xmath13 and @xmath14 , whereas for single - resolved or double - resolved events one or both values are smaller than 1 . differential cross sections as a function of or in regions of are therefore a sensitive probe of the structure of the photon . the internal structure of jets is studied using the jet shape , which is defined as the fractional transverse jet energy contained in a subcone of radius @xmath15 concentric with the jet axis , averaged over all jets of the event sample : @xmath16 with @xmath17 and @xmath18 the total number of jets analysed . both and cone jets are analysed in this way . as proposed in @xcite , only particles assigned to the jet by the jet finders are considered . events entering the jet shape distributions are required to have at least two jets with a transverse energy 3 gev @xmath19 20 gev and a pseudo - rapidity @xmath20 . in figure [ fig : jsh01](a ) the jet shape , @xmath0 , is shown for the algorithm for both @xmath5 and @xmath21 . here and in subsequent figures the total of statistical and systematic uncertainties added in quadrature is shown where larger than the marker size . the inner error bars show the statistical errors . the first sample is dominated by direct photon - photon interactions and hence by quark - initiated jets . as is demonstrated in the figure , jets in this sample are more collimated than for small values of , where the cross - section is dominated by resolved processes and hence has a large contribution from gluon - initiated jets . in both cases the jets become more collimated with increasing transverse energy , as is shown in figure [ fig : jsh01](c ) . there is no significant dependence on the jet pseudo - rapidity ( figure [ fig : jsh01](d ) ) . both phojet @xcite and pythia @xcite give an adequate description of the jet shapes as can be seen in figures [ fig : jsh01](b ) , ( c ) , and ( d ) . the default choices of 1d @xcite for pythia and lo @xcite for phojet are taken . figure [ fig : jsh02 ] compares the shapes of jets defined by the cone algorithm and the algorithm , in each case compared to the shape as obtained from pythia . as for the -jets , the jets defined by the cone algorithm are more collimated in the quark - dominated sample and always become more collimated for increasing transverse energy , while there is no dependence on the jet pseudo - rapidity . the cone - jets are significantly broader than the jets defined by the algorithm at low . with increasing , jets become more collimated and the two jet algorithms give similar results . while the -jets are well described by pythia and phojet , the jet shapes obtained for the cone - jets are somewhat broader than in the data . only the jet algorithm is used for the measurement of the differential di - jet cross - sections . the experimental results are compared to a perturbative qcd calculation at nlo @xcite which uses the parametrisation of the parton distribution functions of the photon @xcite , and was repeated for the kinematic conditions of the present analysis . the renormalisation and factorisation scales are set to the maximum @xmath22 in the event . the calculation was performed in the @xmath23-scheme with five light flavours and @xmath24 mev . the average of the hadronisation corrections estimated by pythia and herwig have been applied to the calculation for this comparison . in the figures described below the shaded band indicates the theoretical uncertainty estimated by the quadratic sum of two contributions : variation of the renormalisation scale by factors of 0.5 and 2 and the difference between using herwig or pythia in estimating the hadronisation corrections . due to the different nature of the underlying partonic process one expects different distributions of the angle @xmath27 between the jet axis and the axis of the incoming partons or direct photons in the di - jet centre - of - mass frame . the leading order direct process @xmath28 proceeds via the @xmath29-channel exchange of a spin-@xmath30 quark , which leads to an angular dependence @xmath31 . in double resolved processes the sum of all matrix elements , including a large contribution from spin-@xmath32 gluon exchange , leads to an approximate angular dependence @xmath33 the contribution of the different processes to all resolved events depends on the parton distribution functions of the photon . an estimator of the angle @xmath27 can be formed from the pseudo - rapidities of the two jets as @xmath34 where it is assumed that the jets are collinear in @xmath35 and have equal transverse energy . only can be measured , as the ordering of the jets in the detector is arbitrary . to obtain an unbiased distribution of the measurement needs to be restricted to the region where the di - jet invariant mass @xmath36 is not influenced by the cuts on @xcite . in the present analysis a cut of @xmath37 gev ensures that the distribution is not biased by the restrictions on for the range @xmath380.8 and @xmath39 confines the measurement to the region where the detector resolution on is good . figure [ fig : costhst ] shows the differential di - jet cross - section as a function of for both @xmath5 and @xmath4 . the steeper rise with increasing from the dominating spin-@xmath32 gluon exchange in the second sample is clearly visible . the shape of both samples is well described by nlo qcd . for @xmath4 the nlo calculation is about 20% below the data . it should be noted that in this region the contribution from the underlying event , not included in the calculation , is expected to be largest , as discussed in more detail below . for @xmath5 the nlo qcd prediction is about 20% above the data . while here the contribution from mia is small , this region is affected by rather large hadronisation corrections , which translates into an uncertainty of the normalisation in comparing the theoretical prediction to the data . probably more importantly it has been pointed out that the calculation of the cross section becomes increasingly problematic when approaching @xmath40 @xcite . the differential di - jet cross - section as a function of the mean transverse energy @xmath26 of the di - jet system is shown in figure [ fig : etmxs ] . at high the cross - section is expected to be dominated by direct processes , associated with the region @xmath5 . consequently we observe a significantly softer spectrum for the case @xmath21 than for the full --space . the calculation is in good agreement with the data for the full --range and for or @xmath41 . the cross - section predicted for @xmath21 is again below the measurement . pythia 6.161 is in good agreement with the measured distributions using the 1d parton densities . the three plots of figure [ fig : xgcomb ] show the differential cross section as a function of for the three regions in --space described above . the shaded histogram on the bottom of each of the three plots indicates the contribution of mia to the cross section as obtained from the pythia @xcite mc generator . it is evident especially for @xmath381 that the mia contribution is of about the same size as the discrepancy between the measurement and the nlo prediction . furthermore it is interesting to observe that there is next to no mia contribution to the cross section if either or is required to be less than one , while the sensitivity to the photon structure at small is retained . as one would expect also the agreement of the nlo calculation with the measurement is best in this case . with these measurements one is therefore able to disentangle the hard subprocess from soft contributions and make the firm statement that nlo perturbative qcd is adequate to describe di - jet production in photon - photon collisions in the regions of phase space where the calculation can be expected to be complete and reliable , i.e. where mia contributions are small and for not too close to unity . at the same time a different sub - set of observables can be used to study in more detail the nature of the soft processes leading to the underlying event .	di - jet production is studied in collisions of quasi - real photons at centre - of - mass energies from 189 to 209 gev at lep . the data were collected with the opal detector . the structure of jets is investigated and differential cross sections are measured and compared to qcd calculations .
nowadays , density functional theory @xcite ( dft ) is considered as a standard method in condensed matter physics , to study electronic , structural and macroscopic properties of solids from first principles . combined with adiabatic perturbation theory , it allows _ a priori _ the computation of derivatives of the energy and related thermodynamic potentials up to any order . at the second order , this approach has been applied to compute linear response functions such as phonon frequencies or born effective charges with an accuracy of a few percents . the third - order derivatives are related to non - linear properties such as phonon lifetimes , raman tensors or non - linear optical susceptibilities . the linear - response formalism has been implemented in various first - principles codes and is routinely applied to an increasing number of systems [ see for example ref . [ ] and references therein ] . by contrast , the non - linear response formalism has been mostly restricted to quantum chemistry problems . although the hyperpolarizabilities of a huge number of molecules have been computed , taking into account both electronic and vibrational ( ionic ) contributions @xcite , applications in condensed matter physics have focused on rather simple cases @xcite . here , we present a methodology for the computation of nonlinear response functions and related physical quantities of periodic solids from density functional perturbation theory ( dfpt ) . we focus on perturbations characterized by a zero wavevector and involving either three electric fields , or two electric fields and one atomic displacement . following nunes and gonze @xcite , our approach makes use of the @xmath1 theorem applied to an electric - field - dependent energy functional @xcite . we report the local density approximation ( lda ) expressions , as implemented within the abinit package @xcite . our paper is organized as follows . in sec . [ sec_form ] , we describe the theoretical background related to the @xmath1 theorem and the electric field perturbation . in sec . [ sec_nlo ] , we describe the computation of the non - linear optical susceptibilities , the non - resonant raman scattering efficiencies of both transverse ( to ) and longitudinal ( lo ) zone - center optical phonons and the linear electrooptic ( eo ) tensor . in sec . [ sec_res ] , we illustrate the validity of the formalism by applying our methodology to some semiconductors and ferroelectric oxides and we briefly discuss the effect of a scissors correction on the eo coefficients and non - linear optical susceptibilities . some of the tensors we consider in this work depend on static electric fields : they include contributions of both the electrons and the ions . other quantities imply only the response of the valence electrons : they are defined for frequencies of the electric field high enough to get rid of the ionic contributions but sufficiently low to avoid electronic excitations . for clarity , we adopt the following convention . static fields will be labeled by greek indices ( @xmath2 , @xmath3 , ... ) while we will refer to optical fields with roman symbols ( @xmath4 , @xmath5 , ... ) . to simplify the notations , we will also drop labels such as `` @xmath6 '' for quantities that do not involve the response of the ions . using this convention , we can write @xmath7 and @xmath8 , respectively , for the optical and static dielectric tensor , respectively , and @xmath9 for the linear eo ( pockels ) tensor that involves two optical and one static electric field . in this section , we present the general framework of the computation of third order energy derivatives based on the @xmath0 theorem @xcite . using the notations of refs . [ ] , we consider three hermitian perturbations labeled @xmath10 , @xmath11 and @xmath12 . the mixed third - order derivatives @xmath13 can be computed from the ground - state and first - order wavefunctions @xmath14 with @xmath15 - \sum_{\alpha,\beta } \lambda_{\beta \alpha}^{\lambda_2 } \langle \psi_{\alpha}^{\lambda_1 } \| \psi_{\beta}^{\lambda_3 } \rangle \nonumber \\ % & & + % \frac{1}{6 } \int d\textbf{r } d\textbf{r } ' d\textbf{r } '' \frac{\delta^3 e_{hxc}[n^{(0 ) } ] } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) \delta n(\textbf{r } '' ) } n^{\lambda_1}(\textbf{r } ) n^{\lambda_2}(\textbf{r } ' ) n^{\lambda_3}(\textbf{r } '' ) \nonumber \\ % & & + % \frac{1}{2 } \int d\textbf{r } d\textbf{r } ' \left . \frac{d}{d \lambda_{2 } } \frac{\delta^2 e_{hxc}[n^{(0 ) } ] } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) } \right \|_{\lambda = 0 } n^{\lambda_1}(\textbf{r } ) n^{\lambda_3}(\textbf{r } ' ) \nonumber \\ % & & + % \frac{1}{2 } \int d\textbf{r } \left . \frac{d^2}{d \lambda_{1 } d \lambda_{3 } } \frac{\delta e_{hxc}[n^{(0 ) } ] } { \delta n(\textbf{r } ) } \right \|_{\lambda = 0 } n^{\lambda_2}(\textbf{r } ) + \left . \frac{1}{6 } \frac{d^3 e_{hxc}[n^{(0 ) } ] } { d \lambda_{1 } d \lambda_{2 } d \lambda_{3 } } \right \|_{\lambda = 0}.\end{aligned}\ ] ] @xmath16 is the kinetic energy and @xmath17 ( resp . @xmath18 ) is the sum of the hartree and exchange - correlation energy ( resp . potential ) . the first - order potential @xmath19 can be computed as a second - order functional derivative of @xmath17 @xcite @xmath20}{\delta n(\bm{r } ) \delta n(\bm{r } ' ) } n^{\lambda_2}(\bm{r } ' ) \ ; d \bm{r } ' + \left . \frac{d}{d \lambda_{2 } } \frac{\delta e_{hxc}[n^{(0)}]}{\delta n(\bm{r } ) } \right \|_{\lambda = 0}.\ ] ] within the parallel gauge , the first - order lagrange multipliers are given by @xmath21 as a consequence of the `` 2n + 1 '' theorem , the evaluation of eq . ( [ eq_3dte_exp ] ) requires no higher order derivative of the wavefunctions than the first one . these first - order wavefunctions are nowadays available in several first - principles codes . they can be computed from linear response by minimizing a stationary expression of the second - order energy @xcite or equivalenty by solving the corresponding sternheimer equation @xcite . it follows that the computation of third - order energies does not require additional quantities than the calculation of second - order energy derivatives . ( [ eq_3dte_exp ] ) is the general expression of the third - order energy derivatives . in case at least one of the perturbations does not affect the explicit form of the kinetic energy or the hartree and exchange - correlation energy , it can be simplified : some of the terms may be zero . this is the case for the electric field perturbations treated in this work as well as for phonon type perturbations . further simplifications can be made in case pseudopotentials without non - linear exchange - correlation core - correction are used . as mentioned in the introduction , special care is required in case one of the perturbations @xmath22 is a macroscopic electric field @xmath23 . in fact , as discussed in the litterature , for infinite periodic solids , usually treated with born - von karman boundary conditions , the scalar potential @xmath24 breaks the periodicity of the crystal lattice . moreover , it is unbound from below : it is always possible to lower the energy by transferring electrons from the valence states to the conduction states in a distant region ( zener breakdown ) . however , for sufficiently small fields , the tunneling current through the band gap can be neglected and the system is well described by a set of electric field dependent wannier functions @xmath25 . as shown by nunes and vanderbilt @xcite , these wannier functions minimize the energy functional @xmath26 = e_0 \left [ w_n \right ] - \omega \bm{\mathcal{e } } \cdot \bm{\mathcal{p}}\ ] ] where @xmath27 is the kohn - sham energy under zero field and @xmath28 the macroscopic polarization that can be computed from the wannier function centers . it is important to note that these wannier functions do not correspond to the true ground - state of the system but rather to a long lived metastable state . in practical applications , it is not mandatory to evaluate the functional eq . ( [ eq_energy ] ) in a wannier basis . it can equivalently be expressed using bloch functions @xmath29 related to @xmath30 by a unitary transform . in this case , the polarization can be computed as a berry phase of the occupied bands @xcite @xmath31 where bz is the brillouin zone , @xmath32 the absolute value of the electronic charge and the factor of 2 accounts for spin degeneracy . the bloch functions are chosen to satisfy the periodic gauge condition @xmath33 in order to use eq . ( [ eq_ksvb ] ) in practical calculations , the integration over the bz , as well as the differentiation with respect to * k * , have to be performed on a discrete mesh of @xmath34 * k-points . in case of the ground - state polarization , the standard approach is to build strings of k-points parallel to a vector of the reciprocal space @xmath35 . the polarization can then be computed as a string - averaged berry phase . unfortunately , the adaptation of this method to the computation of the energy derivatives is plagued with several difficulties , like the following . the general form of the non - linear optical susceptibility tensor of a compound is imposed by its symmetry . for example , in zinc - blende semiconductors , this tensor , expressed in _ cartesian coordinates _ reduces to @xmath36 where @xmath37 is the levi - civita tensor . it follows that the _ reduced coordinates _ formulation of @xmath38 satisfy the relation @xmath39 where at least one of the three indices @xmath4 , @xmath5 and @xmath40 are different from the two others . when we tried to use strings of k-points to compute @xmath38 , eq . ( [ eq_chizb ] ) was not satisfied . however , we were able to avoid these problems , by using the finite difference formula of marzari and vanderbilt @xcite on a regular grid of special k-points ( instead of strings ) @xmath41\ ] ] where b * is a vector connecting a * k-point to one of its nearest neighbours and @xmath42 is a weight factor . the sum in eq . ( [ eq_mv ] ) includes as many shells of first neighbours as necessary to satisfy the condition @xmath43 where @xmath44 are the reduced coordinates of b * and @xmath45 is the metric tensor associated to the real space crystal lattice . in case of the ground - state polarization , we can not apply the discretization eq . ( [ eq_mv ] ) directly to eq . ( [ eq_ksvb ] ) . as shown by marzari and vanderbilt , the discretization of eq . ( [ eq_ksvb ] ) does not transform correctly under the gauge transformation @xmath46 since eq . ( [ eq_gauge ] ) is equivalent to a shift of the origin by one lattice vector @xmath47 , @xmath28 must change accordingly by one polarization quantum . in order to obtain a discrete expression that matches this requirement , we must combine eq . ( [ eq_mv ] ) with the king - smith and vanderbilt formula @xcite @xmath48\ ] ] where @xmath49 is the overlap matrix between bloch functions at * k * and @xmath50 : @xmath51 as discussed by nunes and gonze @xcite , when we apply the perturbation expansion of the preceeding section to the energy functional eq . ( [ eq_energy ] ) , we can adopt two equivalent approaches . the first possibility is the use of eq . ( [ eq_ksvb ] ) for the polarization and a discretization after having performed the perturbation expansion . the second possibility is to apply the @xmath0 theorem directly to eq . ( [ eq_poldisc ] ) in which case no additional discretization is needed . using the notations of nunes and gonze , we will refer to the first case as the dape ( discretization after perturbation expansion ) formulation and to the second one as the pead ( perturbation expansion after discretization ) formulation of the third - order energy . in the following sections , we will discuss both expressions . in addition , in sec . [ sec_res_sc ] , we will compare their convergence with respect to the * k-point sampling on a realistic example . the perturbation expansion of the first term ( @xmath27 ) of eq . ( [ eq_energy ] ) can easily be performed as it is described in the sec . [ sec_der ] . at the opposite , the expansion of the second term ( @xmath52 ) is more tricky since it explicitely depends on the polarization . in the two sections that follow , we will focus on the @xmath52 term of eq . ( [ eq_energy ] ) . it will be referred to as @xmath53 . according to the formalism of the preceeding section , the @xmath54 term acts as an additional external potential that has to be added to the ionic one . the @xmath54 perturbation is linear in the electric field and does not depend explicitely on other variables such as the atomic positions . it just enters the terms of eq . ( [ eq_3dte_exp ] ) that involve the first derivative of @xmath55 with respect to @xmath23 . in other words , the only terms in eq . ( [ eq_perm ] ) that involve the expansion of @xmath28 are of the form @xmath56 where @xmath57 and @xmath58 represent an arbitrary perturbation such as an electric field or an atomic displacement . the dape expression of the third - order derivative of e@xmath59 writes as follows @xmath60 are the projection of the first - order wavefunctions on the conduction bands . the complete expression of various third - order energy derivatives , taking into account the expansion of both e@xmath61 and e@xmath59 , are reported in sec . [ sec_nlo ] . ( [ eq_dape ] ) was derived first by dal corso and mauri @xcite in a slightly different context : they performed the perturbation expansion of the energy functional eq . ( [ eq_energy ] ) using a wannier basis . the resulting expression of the third - order energy was expressed in terms of bloch functions by applying a unitary transform to the wannier orbitals . using the finite difference expression of marzari and vanderbilt eq . ( [ eq_mv ] ) , eq . ( [ eq_dape ] ) becomes @xmath62 where @xmath63 is a basis vector of the reciprocal lattice . applying directly the @xmath0 theorem to eq . ( [ eq_poldisc ] ) we obtain the alternative pead formulation of the third - order energy : @xmath64\end{aligned}\ ] ] where @xmath65 is the inverse of the overlap matrix @xmath49 and @xmath66 its first - order perturbation expansion @xmath67 in the previous section we have discussed the general expressions of third energy derivatives . we now particularize them to the computation of selected non - linear properties . in an insulator the polarization can be expressed as a taylor expansion of the macroscopic electric field @xmath68 where @xmath69 is the zero - field ( spontaneous ) polarization , @xmath70 the linear dielectric susceptibility ( second rank tensor ) and @xmath38 the second - order non - linear optical susceptibility ( third rank tensor ) . in the litterature on non - linear optics , one often finds another definition of the non - linear optical susceptibility : instead of @xmath38 , it is more convenient to rely on the @xmath71-tensor defined as @xmath72 in general , the polarization depends on valence electrons as well as ions . in the present section , we deal only with the electronic contribution : we will consider the ionic cores as clamped to their equilibrum positions . this constraint will be relaxed in the following sections where we allow for ionic displacements . experimentally , the electronic contribution to the linear and non - linear susceptibilities corresponds to measurements for electric fields at frequencies high enough to get rid of the ionic relaxation but low enough to avoid electronic excitations . in case of the second - order susceptibilities , this constraint implies that both the frequency of @xmath23 , and its second harmonic , are lower than the fundamental absorption gap . the general expression of the electronic non - linear optical susceptibility depends on the frequencies of the optical electric fields [ see for example ref . [ ] ] . in the present context of the @xmath0 theorem applied within the lda to ( static ) dft , we neglect the dispersion of @xmath38 . as a consequence , @xmath38 satisfies kleinman s @xcite symmetry condition which means that it is symmetric under a permutation of @xmath4 , @xmath5 and @xmath40 . in order to be able to investigate its frequency dependence , one would need either to apply the formalism of time - dependent dft @xcite or to use expressions that involve sums over excited states @xcite . following the work of dal corso and co - workers @xcite we can relate the non - linear optical susceptibilities to a third - order derivative of the energy with respect to an electric field @xmath73 where @xmath74 is defined as the sum over the permutations of the three perturbations @xmath75 ( [ eq_perm ] ) . using the pead formulation of sec . [ sec_elfd ] we can express these terms as follows @xmath76 \nonumber \\ % & & % + \frac{2}{n_k } \sum_{\bm{k } } \sum_{n , m}^{occ } \bigglb [ \delta_{m , n } \langle u_{n \textbf{k}}^{\mathcal{e}_{i } } \| v_{hxc}^{\mathcal{e}_{j } } - \langle u_{m \textbf{k}}^{(0 ) } \| v_{hxc}^{\mathcal{e}_{j } } \langle u_{n \textbf{k}}^{\mathcal{e}_{i } } \| u_{m \textbf{k}}^{\mathcal{e}_{l } } \rangle \bigglb ] \nonumber \\ % & & + % \frac{1}{6 } \int d\textbf{r } d\textbf{r } ' d\textbf{r } '' \frac{\delta^3 e_{xc}[n^0 ] } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) \delta n(\textbf{r } '' ) } n^{\mathcal{e}_i}(\textbf{r } ) n^{\mathcal{e}_j}(\textbf{r } ' ) n^{\mathcal{e}_l}(\textbf{r}'').\end{aligned}\ ] ] we now consider the computation of raman scattering efficiencies of zone - center optical phonons . in the limit @xmath77 , the matrix of interatomic force constants @xmath78 can be expressed as the sum of an analytical part and a non - analytical term @xcite @xmath79 the analytical part corresponds to the second - order derivative of the energy with respect to an atomic displacement at @xmath80 under the condition of vanishing macroscopic electric field . the second term is due to the long - range electrostatic interactions in polar crystals . it is at the origin of the so - called lo - to splitting and can be computed from the knowledge of the born effective charges @xmath81 and the electronic dielectric tensor @xcite @xmath7 . the phonon frequencies @xmath82 and eigendisplacements @xmath83 are solution of the following generalized eigenvalue problem @xmath84 where @xmath85 is the mass of atom @xmath86 . as a convention , we choose the eigendisplacements to be normalized as @xmath87 in what follows we consider non - resonant raman scattering where an incoming photon of frequency @xmath88 and polarization @xmath89 is scattered to an outgoing photon of frequency @xmath90 and polarization @xmath91 by creating a phonon of frequency @xmath82 ( stokes process ) . the scattering efficiency @xcite ( cgs units ) corresponds to @xmath92 where @xmath93 is the speed of light in vacuum and @xmath94 the boson factor @xmath95 the raman susceptibility @xmath96 is defined as @xmath97 where @xmath98 is the electronic linear dielectric susceptibility tensor . @xmath99 is the angle of collection in which the outgoing photon is scattered . due to snell s law , @xmath99 is modified at the interface between the sample and the surrounding medium . experimentally , the scattering efficiencies are measured with respect to the solid angle of the medium while eq . ( [ eq_efficiency ] ) refers to the solid angle inside the sample . in order to relate theory and experiment , one has to take into account the different refractive indices of the sample and medium . for example , in the isotropic case , eq . ( [ eq_efficiency ] ) has to be multiplied @xcite by @xmath100 where @xmath101 and @xmath102 are respectively the refractive indices of the sample and the medium . for pure transverse optical phonons , @xmath103 can be computed as a mixed third - order derivative of the energy with respect to an electric field , twice , and to an atomic displacement under the condition of zero electric field @xmath104 in case of longitudinal optical phonons with wavevector @xmath77 in a polar crystal , eq . ( [ eq_ramansus ] ) must take into account the effect of the macroscopic electric field generated by the lattice polar vibration . this field enters the computation of the raman susceptibilities at two levels . on one hand , it gives rise to the non - analytical part of the matrix of interatomic force constants eq . ( [ eq_ifc ] ) that modifies the frequencies and eigenvectors with respect to pure transverse phonons . on the other hand , the electric field induces an additional change in the dielectric susceptibility tensor related to the non - linear optical coefficients @xmath105 . for longitudinal optical phonons , eq . ( [ eq_dchidtau ] ) has to be modified as follows @xcite : @xmath106 the mixed third - order derivatives ( [ eq_dchidtau ] ) can be computed from various techniques including finite differences of the dielectric tensor @xcite or the second derivative of the electronic density matrix @xcite . here , we follow an approach similar to deinzer and strauch @xcite based on the @xmath1 theorem . the third - order energy can be computed as the sum over the 6 permutations eq . ( [ eq_perm ] ) of @xmath107 , @xmath108 and @xmath109 . according to the discussion of sec . [ sec_elfd ] , we have to distinguish between the terms that involve the discretization of the polarization such as @xmath110 or @xmath111 and those that can be computed from a straightforward application of the @xmath1 theorem such as @xmath112 . the former ones show an electric field as second perturbation . they can be computed from an expression analogous to eq . ( [ eq_e3_elfd ] ) @xmath113 \nonumber \\ % & & % + \frac{2}{n_k } \sum_{\bm{k } } \sum_{n , m}^{occ } \bigglb [ \delta_{m , n } \langle u_{n \textbf{k}}^{\tau_{\kappa \lambda } } \| v_{hxc}^{\mathcal{e}_{i } } - \langle u_{m \textbf{k}}^{(0 ) } \| v_{hxc}^{\mathcal{e}_{i } } \langle u_{n \textbf{k}}^{\tau_{\kappa \lambda } } \| u_{m \textbf{k}}^{\mathcal{e}_{j } } \rangle \bigglb ] \nonumber \\ % & & + % \frac{1}{6 } \int d\textbf{r } d\textbf{r } ' d\textbf{r } '' \frac{\delta^3 e_{xc}[n^0 ] } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) \delta n(\textbf{r } '' ) } n^{\tau_{\kappa \lambda}}(\textbf{r } ) n^{\mathcal{e}_{i}}(\textbf{r } ' ) n^{\mathcal{e}_{j}}(\textbf{r}'').\end{aligned}\ ] ] we obtain a similar expression for @xmath111 . the remaining terms do not require any differentiation with respect to k. they can be computed from the first - order change of the ionic ( pseudo ) potential with respect to an atomic displacement @xmath114 @xmath115 \nonumber \\ % & & % + \frac{1}{2 } \int d\textbf{r } d\textbf{r } ' \left . \frac{d}{d \tau_{\kappa \lambda } } \frac{\delta^2 e_{hxc } } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) } \right \|_{n^{(0 ) } } n^{\mathcal{e}_{i}}(\textbf{r } ) n^{\mathcal{e}_{j}}(\textbf{r } ' ) \nonumber \\ % & & % \frac{1}{6 } \int d\textbf{r } d\textbf{r } ' d\textbf{r } '' \frac{\delta^3 e_{xc}[n^0 ] } { \delta n(\textbf{r } ) \delta n(\textbf{r } ' ) \delta n(\textbf{r } '' ) } n^{\tau_{\kappa \lambda}}(\textbf{r } ) n^{\mathcal{e}_{i}}(\textbf{r } ' ) n^{\mathcal{e}_{j}}(\textbf{r}'').\end{aligned}\ ] ] in pseudopotential calculations , the computation of the first - order ionic potential @xmath114 requires the derivative of local and non - local ( usually separable ) operators . these operations can be performed easily without any additional workload as described in ref . [ ] . in spite of the similarities between eqs . ( [ eq_e3_tee ] ) and ( [ eq_e3_ete ] ) and the expression proposed by deinzer and strauch we can quote few differences . first , our expression of the third - order energy makes use of the pead fomulation for the expansion of the polarization . moreover , eq . ( [ eq_e3_ete ] ) is more general since it allows the use of pseudopotentials with non - linear core correction through the derivative of the second - order exchange - correlation energy with respect to @xmath107 ( third term ) . as in the cases of the born effective charges and of the dynamical matrix @xcite , the coefficients @xmath116 must vanish when they are summed over all atoms in the unit cell . @xmath117 physically , this sum rule guarantees the fact that the macroscopic dielectric susceptibility remains invariant under a rigid translation of the crystal . in practical calculations , it is not always satisfied although the violation is generally less severe than in case of @xmath78 or @xmath118 . even in calculations that present a low degree of convergence , the deviations from this law can be quite weak . they can be corrected using similar techniques as in case of the born effective charges @xcite . for example , we can define the mean excess of @xmath116 per atom @xmath119 and redistribute it equally between the atoms @xmath120 the optical properties of a compound usually depend on external parameters such as the temperature , electric fields or mechanical constraints ( stress , strain ) . in the present section we consider the variations of the refractive index induced by a static or low - frequency electric field @xmath121 . at the linear order , these variations are described by the linear eo coefficients ( pockels effect ) @xmath122 where @xmath123 is the inverse of the electronic dielectric tensor and @xmath9 the eo tensor . within the born and oppenheimer approximation , the eo tensor can be expressed as the sum of three contributions : a bare electronic part @xmath124 , an ionic contribution @xmath125 and a piezoelectric contribution @xmath126 . the electronic part is due to an interaction of @xmath121 with the valence electrons when considering the ions artificially as clamped at their equilibrium positions . it can be computed from the non - linear optical coefficients . as can be be seen from eq . ( [ eq_nlo_pol ] ) , @xmath127 defines the second - order change of the induced polarization with respect to @xmath121 . taking the derivative of eq . ( [ eq_nlo_pol ] ) , we also see that @xmath127 defines the first - order change of the linear dielectric susceptibility , which is equal to @xmath128 since the eo tensor depends on @xmath129 rather than @xmath130 , we have to transform @xmath130 by the inverse of the zero field electronic dielectric tensor @xcite @xmath131 using eq . ( [ eq_transfo ] ) we obtain the following expression for the electronic eo tensor @xmath132 eq . ( [ eq_eo_el ] ) takes a simpler form when expressed in the principal axes of the crystal under investigation @xcite @xmath133 where the @xmath134 coefficients are the principal refractive indices . the origin of the ionic contribution to the eo tensor is the relaxation of the atomic positions due to the applied electric field @xmath121 and the variations of @xmath7 induced by these displacements . it can be computed from the born effective charges @xmath135 and the @xmath136 coefficients introduced in sec . [ sec_raman ] . as shown in appendix [ sec_app ] [ see also refs . ] , the ionic eo tensor can be computed as a sum over the transverse optic phonon modes at @xmath80 @xmath137 where @xmath96 is the raman susceptibility of mode @xmath138 [ eq . ( [ eq_ramansus ] ) ] and @xmath139 the mode polarity @xmath140 that is directly linked to the modes oscillator strength @xmath141 for simplicity , we have expressed eq . ( [ eq_eo_ion ] ) in the principal axes while a more general expression can be derived from eq . ( [ eq_transfo ] ) . finally , the piezoelectric contribution is due to a relaxation of the unit cell shape due to the converse piezoelectric effect @xcite . as it is discussed in appendix [ sec_app ] , it can be computed from the elasto - optic coefficients @xmath142 and the piezoelectric strain coefficients @xmath143 @xmath144 in the discussion of the eo effect , we have to specify whether we are dealing with strain - free ( clamped ) or stress - free ( unclamped ) mechanical boundary conditions . the clamped eo tensor @xmath145 takes into account the electronic and ionic contributions but neglects any modification of the unit cell shape due to the converse piezoelectric effect @xcite @xmath146 experimentally , it can be measured for frequencies of @xmath121 high enough to eliminate the relaxations of the crystal lattice but low enough to avoid excitations of optical phonon modes ( usually above @xmath147 100 mhz ) . to compute the unclamped eo tensor @xmath148 , we have to add the piezoelectric contribution to @xmath145 @xmath149 experimentally , @xmath148 can be measured for frequencies of @xmath121 below the ( geometry dependent ) mechanical body resonances of the sample @xcite ( usually below @xmath147 1 mhz ) . our calculations have been performed within the local density approximation ( lda ) to the density functional theory @xcite ( dft ) . we used the abinit @xcite package , a planewave , pseudopotential dft code @xcite in which we have implemented the formalism presented above . for reasons that will become obvious below , we chose the pead formulation eq . ( [ eq_pead ] ) to perform the differentiation with respect to k. for the exchange - correlation energy @xmath150 we relied on the parametrization of perdew and wang @xcite as well as the parametrization of goedecker , teter and hutter @xcite . these expressions have the advantage to avoid any discontinuities in the functional derivative of @xmath150 . in case of the semiconductors si , alas and alp , we used at @xmath151 grid of special k - points , a plane - wave kinetic energy cutoff of 10 hartree and troullier - martins @xcite norm - conserving pseudopotentials . these calculations have been performed at the theoretical lattice constant . to perform the finite difference calculations of the raman polarizabilities , changes of the electronic dielectric tensor were computed for atoms displaced by @xmath152 of the unit cell parameter along the cartesian directions . in case of rhombohedral batio@xmath153 , we used a @xmath154 grid of special k - points , a plane - wave kinetic energy cutoff of 45 hartree and extended norm - conserving pseudopotentials @xcite . since the ferroelectric instability is quite sensitive to the volume of the unit cell and tends to disappear due to the volume underestimation of the lda @xcite , we chose to work at the experimental lattice constants . at the opposite to the lattice parameters , the atomic positions have been relaxed : the residual forces on the atoms were smaller than @xmath155 hartree / bohr . it was shown by gonze , ghosez and godby @xcite that an accurate functional for the exchange - correlation energy in extended systems should depend on both the density and the polarization . the lda used here neglects this polarization dependence and may consequently indroduce significant relative errors when studying the response of a solid to an electric field . in case of the second - order derivatives , the lda usually yields an overestimate of the dielectric tensor ( as large as 20 % in batio@xmath156 ) @xcite . at the opposite , no clear trends have been reported yet concerning non - linear optical properties such as @xmath127 . @xcite in lda calculations , it is common practice to apply a _ scissors correction _ @xcite to compensate the lack of polarization dependence of the exchange - correlation functional . in case of non - linear optical properties , such a correction can be applied at different levels . on the one hand , we can compute the non - linear optical susceptibilities ( eq . ( [ eq_e3_elfd ] ) ) using a scissors operator for the first - order wavefunctions @xcite . on the other hand , in the computation of the eo coefficients , we can use a scissors corrected refractive index in eqs ( [ eq_eo_el1 ] ) and ( [ eq_eo_ion ] ) . the influence of these corrections will be discussed below . in order to illustrate the computation of third - order energy derivatives , we performed a series of calculations on various cubic semiconductors . in these compounds , the non - linear optical susceptibility tensor @xmath157 and the raman susceptibility tensor @xmath158 only have one independent element d@xmath159 and @xmath160 . also , instead of @xmath160 it is customary to report the raman polarizability @xcite defined as @xmath161 where @xmath162 is the reduced mass of the two atoms in the unit cell . the formalism of sec . [ sec_form ] involves an integration over the bz and a differentiation with respect to k. in practical calculations , these operations must be performed on a discrete mesh of special k-points . as we explained in sec . [ sec_form ] , the discretization can either be performed before ( pead ) or after ( dape ) the perturbation expansion of the energy functional eq . ( [ eq_energy ] ) . up to know , the applications of the present formalism to real materials @xcite made use of the dape formula of the third - order energy . the only application of the pead formula has been reported by nunes and gonze @xcite on a one - dimensional model system . these authors observed that the pead formula converges better with respect to the k-point sampling than the dape formula . in order to compare the performance of these two approaches on a realistic case , we applied both of them to compute the non - linear optical susceptibility d@xmath159 of alas . we performed a series of calculations on a @xmath163 grid of special k-points . as can be seen on figure [ fig_convergence ] the pead formula converges much faster than the dape formula . therefore , the pead formulation has been applied to obtain the results presented below . it is the one that is actually available in the abinit code . non - linear optical susceptibility d@xmath159 ( pm / v ) of alas for various grids of @xmath163 special k-points.,width=226 ] in table [ tab_scnl ] , we report the non - linear optical susceptibilities of the cubic semiconductors alas and alp . our results are in close agreement with the values obtained by dal corso and co - workers @xcite who applied the @xmath0 theorem within the dape formalism , the results of levine and co - workers @xcite who used a `` sum over excited states '' technique and the values obtained by souza and co - workers @xcite who followed a finite electric field approach . the values in the lower part of table [ tab_scnl ] have been obtained using a scissors correction . our methodology provides a correction similar to what is reported by levine and allan @xcite . the scissors correction decreases the value of the non - linear optical susceptibilities in agreement with the discussion of ref . . to the authors knowledge , no experimental data are available for alas and alp . for other cubic semiconductors , it is however not clear that the use of a scissors correction improves the agreement with the experiment @xcite and will even have a negative effect when the lda underestimates the experimental value . in addition , it is not straightforward to isolate the error of the lda on the non - linear response functions from other sources of errors . other factors have a similar strong influend on @xmath127 as the scissors correction . for example , the values of the non - linear optical susceptibilities strongly depend on the pseudopotential @xcite or on the error on the unit cell volume @xcite that is usually underestimated in lda calculations . .[tab_scnl ] non - linear optical susceptibilities @xmath164 ( pm / v ) of some semiconductors . the values in the lower part of the table have been obtained using a scissors ( sci ) correction . [ cols="<,<,^,^",options="header " , ] in this paper , we presented the general framework for the computation of third - order energy derivatives within dft . our formalism makes use of the @xmath165 theorem and the modern theory of polarization . focusing on derivatives that are characterized by a zero wavevector and that involve either three electric fields or two electric fields and one atomic displacement , we described the computation of non - linear optical susceptibilities , of raman scattering efficiencies of to and lo phonons and of the eo tensor . the computation of the berry phase polarization involves a derivative of the wavefunctions with respect to their wavevector . in practice , this differentiation is computed on a grid of special k-points . the perturbation expansion can either be performed before ( dape ) or after ( pead ) the discretization , leading to two mathematically distinct expression of the third - order energies . we used both of them to compute the non - linear optical susceptibility of alas and we have shown that the pead formulation converges faster with respect to the k-point sampling . we have computed the non - linear optical susceptibilities and raman polarizabilities of some cubic semiconductors as well as the eo tensor in the rhombohedral phase of batio@xmath153 . finally , we have studied the effect of a scissors correction on the eo coefficients and the non - linear optical susceptibilities . at the opposite to the dielectric tensor , we did not find a systematic improvement of the results by using this correction . we can figure out several applications of the methodology presented in this work . combined with the calculation of phonon frequencies and infrared intensities , the computation of raman efficiencies can be a useful complementary tool for the interpretation of experimental spectra . furthermore , the computation of the eo tensor from first - principles can guide the tuning of the eo properties and help designing new efficient eo materials . this could reveal particularly helpful since accurate optical measurements require high quality single crystals not always directly accessible . the authors are grateful to m. d. fontana , p. bourson , b. kirtman and b. champagne for helpful discussions . m. v. and x. g. acknowledge financial support from the fnrs belgium . this work was supported by the volkswagen - stiftung within the project nano - sized ferroelectric hybrids `` ( i/77 737 ) , the region wallonne ( nomade , project 115012 ) , the communaut francaise de belgique - actions de recherche concertes , the pai / uiap phase 5 ' ' quantum size effects in nanostructured materials `` , fnrs - belgium through grants 9.4539.00 and 2.4562.03 , the european union through the research and training network ' ' exciting " ( hprm - ct-2002 - 00317 ) . as it is discussed in sec . [ sec_eo ] , the optical properties of a compound are modified by an electric field @xmath121 or a mechanical constraint ( a stress @xmath166 or a homogeneous strain @xmath167 ) . at the linear order , the variations of @xmath168 can be described using either the variables @xmath169 or @xmath170 @xcite where @xmath145 and @xmath148 are resp . the clamped ( strain free ) and unclamped ( stress free ) eo coefficients , @xmath142 are the elasto - optic ( strain - optic ) coefficients and @xmath172 are the piezo - optical ( stress - optical ) coefficients . in order to relate eqs . ( [ eq_reta ] ) and ( [ eq_rsigma ] ) , we can express the strain as beeing induced by the stress or by the electric field ( converse piezoelectric effect ) @xmath173 where @xmath174 are the elastic compliances and @xmath143 the piezoelectric strain coefficients . if we assume for example that the unit cell is free to relax within the electric field ( stress - free mechanical boundary conditions ) we can either use eq . ( [ eq_rsigma ] ) ( in which case the second term of the right - hand - side is zero ) or eq . ( [ eq_reta ] ) to compute @xmath175 . in the latter case , the strain induced by the electric field can be obtained from the second term of the right - hand - side of eq . ( [ eq_strain ] ) @xmath176 using this identity , we obtain the following relation between the unclamped and the clamped eo coefficients @xmath177 in order to derive the expressions of the clamped and unclamped eo tensor of sec . [ sec_eo ] , we use a taylor expansion of the electric enthalpy @xcite @xmath178 . similar developments have already been applied to determine the lattice contribution of the static dielectric tensor and of the piezoelectric tensor @xcite . they are based on an expansion of @xmath178 up to the second order in the atomic coordinates @xmath179 , the homogeneous strain @xmath167 and the macroscopic electric field @xmath121 . in this section , we extend these developments to the third order . the electric enthalpy of a solid in an electric field is obtained by the minimization @xmath180 we denote @xmath181 , @xmath182 the atomic positions and the strain that minimize @xmath178 at constant @xmath23 and @xmath183 , @xmath184 (= 0 ) their values at @xmath185 . for small fields , we can expand the function @xmath186 in powers of @xmath23 around @xmath185 : @xmath187 where @xmath99 is the volume of the primitive unit cell in real space and @xmath188 , @xmath189 and @xmath190 are the macroscopic polarization , electronic dielectric tensor and non - linear optical coefficients at zero macroscopic electric field and for a given configuration ( @xmath191 , @xmath192 ) . at non - zero field , these quantities are defined as partial derivatives of @xmath178 with respect to @xmath23 . for example , the electric field dependent electronic dielectric tensor can be computed from the expression @xmath193 let @xmath194 be the displacement of atom @xmath86 along direction @xmath2 and @xmath195 ( resp . @xmath196 ) the first - order modification of the atomic position ( resp . strain ) induced by a perturbation @xmath197 @xmath198 in the discussion that follows , we will study the effect of an electric field perturbation and a strain perturbation on the electric enthalpy @xmath178 in order to obtain the formulas to compute the elasto - optic coefficients as well as the clamped and the unclamped eo tensors . the elasto - optic tensor can be computed from the _ total _ derivative of the dielectric tensor with respect to @xmath167 at zero electric field @xmath200 the derivative in the first term of the right hand side is computed considering the ionic cores as artificially clamped at their equilibrum positions . the remaining terms represent the ionic contribution to the elasto - optic tensor . they involve derivatives of the linear dielectric susceptibility @xmath98 with respect to the atomic positions that have to be multiplied by the first - order strain induced atomic displacements @xmath201 [ eq . ( [ eq_tau ] ) ] . to compute these quantities we use the fact that @xmath178 is minimum at the equilibrum for an imposed strain @xmath192 . this condition implies @xmath202 since we are interested in first - order atomic displacements we can write @xmath203 solving the extremum equation ( [ eq_tau_eq ] ) to the linear order in @xmath192 , we obtain @xmath204 the second derivatives on the left side of eq . ( [ eq_tau_disp ] ) define the matrix of interatomic force constants at zero macroscopic electric field which enables the computation of the transverse phonon frequencies @xmath82 and eigendisplacements @xmath83 . by decomposing @xmath201 in the basis of the zone - center phonon - mode eigendisplacements @xmath205 and using eqs . ( [ eq_eigen ] ) , ( [ eq_norm ] ) we derive the following expression for the first - order strain induced atomic displacements @xmath206 where @xmath207 if we introduce eqs . ( [ eq_ion_decomp ] ) and ( [ eq_tau_disp1 ] ) into eq . ( [ eq_depsdeta ] ) and use the definition of the raman susceptibility eq . ( [ eq_ramansus ] ) and the transformation eq . ( [ eq_transfo ] ) , we finally obtain the formula to compute the elastooptic tensor @xmath208 to simplify , we write eq . ( [ eq_elastoopt ] ) in the principal axes of the crystal under investigation . a more general expression can be obtained from eq . ( [ eq_transfo ] ) . ( [ eq_elastoopt ] ) is different from the approach used previously by detraux and gonze to study the elasto - optic tensor in @xmath2-quartz @xcite . the authors of ref . used finite differences with respect to strains to compute the the total derivative of @xmath7 . in their approach , the atoms where relaxed to their equilibrum positions in the strained configurations . in case of eq . ( [ eq_elastoopt ] ) , the first term of the right - hand - side is computed at clamped atomic positions while the effect of the strain - induced atomic relaxations is taken into account by the second term . the clamped eo tensor can be computed from the _ total _ derivative of the electric field dependent dielectric tensor eq . ( [ eq_ion_diel ] ) with respect to @xmath210 @xmath211 the derivative in the first term is computed considering the ionic cores as artificially clamped at their equilibrum positions . this term represents the bare electronic contribution to the eo tensor that can be computed from the non - linear optical coefficients @xmath212 related to a third - order partial derivative of @xmath178 @xmath213 the remaining terms in eq . ( [ eq_clamp_chidc ] ) represent the ionic contribution to the eo tensor . they involve derivatives of the linear dielectric susceptibility @xmath98 with respect to the atomic positions that have to be multiplied by the first - order electric field induced atomic displacements @xmath214 [ eq . ( [ eq_tau ] ) ] . to obtain these quantities , we proceed the same way as in case of the elasto - optic tensor . using the equilibrum condition @xmath215 and expanding @xmath216 to the first - order in the electric field , we obtain @xmath217 this expression is similar to eq . ( [ eq_tau_disp ] ) . the second - order derivatives of @xmath178 on the left side are the interatomic force constatnts and the derivative of the zero field polarization with respect to @xmath216 on the right side is the born effective charge tensor @xmath218 of atom @xmath86 . decomposing @xmath214 in the basis of the zone - center phonon - mode eigendisplacements [ eq . ( [ eq_ion_decomp ] ) ] and using the orthononormality constraint eq . ( [ eq_norm ] ) we derive the following expression for the first - order electric field induced atomic displacements @xmath219 if we introduce eqs . ( [ eq_elec_ent ] ) and ( [ eq_clamp_disp1 ] ) into eq . ( [ eq_clamp_chidc ] ) we finally obtain the formula to compute the total derivative of the dielectric tensor @xmath220 using the definition of the raman susceptibility [ eq . ( [ eq_ramansus ] ) ] , the mode polarity [ eq . ( [ eq_eo_pol ] ) ] and the transformation [ eq . ( [ eq_transfo ] ) ] we obtain the expression of the clamped eo tensor @xmath221 as in case of the elasto - optic tensor [ eq . ( [ eq_elastoopt ] ) ] , we have written eq . ( [ eq_clamp_rijgamma ] ) in the principal axes of the crystal under investigation . in order to compute the unclamped eo tensor , we have to take into account both the electric field induced atomic displacments @xmath214 and the electric field induced strain @xmath223 when computing the total derivative of @xmath7 @xmath224 the electronic contribution [ first term of eq . ( [ eq_unclamp_chidc ] ) ] is the same as for the clamped eo tensor . it can be computed from the non - linear optical coefficients [ eq . ( [ eq_elec_ent ] ) ] . to compute @xmath214 and @xmath223 , we can use an equilibrum condition similar to eq . ( [ eq_clamp_eq ] ) where we require that the first - order derivatives of @xmath178 with respect to @xmath216 and @xmath167 vanish . expanding @xmath216 and @xmath167 to the first - order in the electric field , we obtain the system of coupled equations [ see also ref . ] because of the coupling between @xmath214 and @xmath223 , defined by the mixed second - order derivatives @xmath226 the second term of the right hand side of eq . ( [ eq_unclamp_chidc ] ) is different from that of eq . ( [ eq_clamp_chidc ] ) . that means that the sum of the first and second term of eq . ( [ eq_unclamp_chidc ] ) is not identical to the clamped eo coefficients @xmath145 . moreover , the third term of eq . ( [ eq_unclamp_chidc ] ) is different from the piezoelectric contribution of sec . [ sec_app_mac ] . in order to obtain the decomposition of @xmath148 into electronic , ionic and piezoelectric contributions defined previously , we can solve eq . ( [ eq_taueta1 ] ) for @xmath214 . in the basis of the zone - center phonon mode eigendisplacements we can write @xmath227 if we insert this relation into eq . ( [ eq_unclamp_chidc ] ) and use the transformation eq . ( [ eq_transfo ] ) we obtain the following expression of the unclamped eo tensor in the principal axes @xmath228 \eta_{\mu \nu}^{\mathcal{e}_{\gamma}}\end{aligned}\ ] ] the sum of the first and second term of the right - hand side of eq . ( [ eq_unclamp_1 ] ) is equal to the clamped eo coefficient @xmath145 . the product of the conversion factor times the bracket in the third term of eq . ( [ eq_unclamp_1 ] ) is equal to the elasto - optic coefficient @xmath142 [ eq . ( [ eq_elastoopt ] ) ] . finally , by definition of the converse piezoelectric effect , @xmath223 is equal to the piezoelectric strain coefficient @xmath143 . we obtain thus the following expression of the unclamped eo coefficients that is equal to the one derived in sec . [ sec_app_mac ] from pure macroscopic arguments @xmath177 it is worth noticing that the so - called piezoelectric contribution does not only take into account the change of the linear optical susceptibility with strain ( third term of the right hand side of eq . ( [ eq_unclamp_chidc ] ) ) but also includes the modification of the ionic contribution , with respect to the clamped case , that is associated to the modification of the ionic relaxation induced by the strain . x. gonze , j .- m . beuken , r. caracas , f. detraux , m. fuchs , g .- m . rignanese , l. sindic , m. verstraete , g. zerah , f. jollet , m. torrent , a. roy , m. mikami , ph . ghosez , j .- y . raty and d.c . allan , computational materials science 25 * , 478 ( 2002 ) . [ url www.abinit.org ]	the non - linear response of infinite periodic solids to homogenous electric fields and collective atomic displacements is discussed in the framework of density functional perturbation theory . the approach is based on the @xmath0 theorem applied to an electric - field - dependent energy functional . we report the expressions for the calculation of the non - linear optical susceptibilities , raman scattering efficiencies and electrooptic coefficients . different formulations of third - order energy derivatives are examined and their convergence with respect to the * k*-point sampling is discussed . we apply our method to a few simple cases and compare our results to those obtained with distinct techniques . finally , we discuss the effect of a scissors correction on the eo coefficients and non - linear optical susceptibilities .
over the past few years we have been witness to vigorous experimental , phenomenological and theoretical investigations of the proton deep inelastic scattering ( dis ) structure functions and some exclusive channels with small @xmath10 and @xmath11 . these comprehensive studies aim at establishing the applicability and possible need for a re - formulation of pqcd , as we know it , when approaching the kinematic interface with the less understood npqcd dominated domain . the standard procedure for the pqcd analysis of dis on a nucleon target has been to utilize the dglap evolution equations for the structure functions as the key ingredient for fixing the parton distribution functions ( p.d.f . ) . these p.d.f . are then used as input for the calculations of exclusive dis channels , usually executed in the color dipole approximation . the physics of small @xmath2 and small @xmath3 is associated with the search for the scale of gluon saturation implied by s - channel unitarity @xcite . one should remember , though , that gluon saturation signals the transition from perturbative to non perturbative qcd . this transition is preceded by sc signatures which are expected to be experimentally visible even though the relevant scattering amplitude has not yet reached the unitarity black limit . moreover , from our experience with soft pomeron physics , we know that different channels have different scales at which unitarity corrections become appreciable . specifically , the scale associated with the diffractive channels are considerably smaller than those associated with the elastic channel @xcite . inspite of significant theoretical progress in recent years @xcite , it is still not clear what the saturation scale in the present experimentally accesible kinematic region is . we recall that , while the global analysis of @xmath1 ( or @xmath12 ) data shows no conclusive deviations from dglap , there are dedicated hera investigations suggesting deviations from the dglap expectations in the small @xmath2 and @xmath3 limits . these signatures are observed in both the fine details of @xmath1 as well as in the diffractive channels , provided @xmath2 and @xmath3 are small enough . in the following we present a detailed study of @xmath4 which is supported by a coupled analysis of @xmath5 photo and dis production . the strategy of our investigation is based on the observation that these observables ( in lla of pqcd ) are linear in @xmath6 and @xmath13 respectively , and being relatively well measured may serve as effective discriminators when we compare their detailed features with existing relevant theoretical approaches and models . in the first phase of our investigation , we found that none of the latest p.d.f . @xcite@xcite@xcite provides an adequate simultaneous reproduction , at small pqcd scales , of the recent hera data on the logarithmic slope of @xmath7 @xcite@xcite at small @xmath2 and @xmath3 , as well as the abundant high energy data on @xmath5 photo and dis production @xcite@xcite@xcite . we then proceed to show that when grv98nlo is corrected for sc , as suggested in our previous publications @xcite@xcite , it gives a good description of these data . the sc calculation depends on one parameter , @xmath8 , which is directly deduced from the @xmath5 photoproduction forward differential slope . we elaborate on a recent suggestion @xcite that the gluon saturation scale may be determined by examining the behaviour of @xmath14 against @xmath2 and @xmath3 at fixed w values . in our opinion , this suggestion which was discussed also in a recent presentation of the zeus new data @xcite , reflects the particular kinematic relationship between @xmath15 and @xmath3 and does not provide an unambiguous determination of the desired scale . our approach can be tested in the high energy analysis of photo and dis exclusive production of @xmath5 . this is seemingly a straight forward procedure as the lla pqcd calculation of this cross section is proportional to @xmath16 , where @xmath3 and @xmath17 are determined from @xmath18 , the c - quark mass , and w , the c.m . a realistic calculation depends on a few corrections to the original pqcd estimate : @xmath19 due to the amplitude real part , @xmath20 due to the contribution of the off diagonal ( skewed ) gluon distribution @xcite and @xmath21 due to the fermi motion deviations from the simple static non relativistic estimate of the @xmath5 wave function @xcite . all of these corrections contribute to the final ( amplitude squared ) estimate . we present a detailed analysis of this channel which is consistent with our approach and input choices made in the analysis of @xmath4 . we conclude with some general comments and observations on the outstanding problems of gluon saturation and screening corrections . checking for unitarity corrections in @xmath7 studies is not simple . as is well known , a global dglap analysis of the data with the recent p.d.f . @xcite@xcite@xcite is adequate . a study @xcite comparing the screened and non screened dglap calculations of @xmath1 showed only a small difference due to sc even in the small @xmath2 and @xmath3 attained by present hera measurements . clearly , a unitarity study in the above kinematic limit requires a dedicated investigation , confined to small @xmath2 and @xmath3 , which can magnify the presumed experimental signatures . we recall that in the small @xmath3 limit of dglap we have [ a ] = xg^dglap(x , q^2 ) . accordingly , a significant deviation of the data from , where @xmath22 is obtained from the global @xmath7 analysis , may serve as an experimental signature indicating the growing importance of unitarity corrections . this was first suggested by caldwell @xcite , showing a rather complicated plot of @xmath4 in which each point had different @xmath2 and @xmath3 values . the caldwell plot suggested a dramatic turn over of @xmath23 corresponding to @xmath2 of about @xmath24 and @xmath25 in contrast to the behavior expected from grv94 @xcite at sufficiently small @xmath2 and @xmath3 . the problem with this presentation is that , as suggestive as it may seem , it does not discriminate between different dynamical interpretations @xcite@xcite@xcite@xcite@xcite . it is actually compatible with an overall data generator @xcite as well as the latest p.d.f . which were re - adjusted to account for this observation . we conclude that , where as a reproduction of the cadwell plot is required as a pre condition for a serious consideration of any suggested model , the plot on its own can not serve as an effective discriminator between models even in the extreme case when they are fundamentally different . a far better discrimination is obtained if we carefully study the small @xmath2 and @xmath3 dependences of @xmath23 at either fixed @xmath2 or fixed @xmath3 values being free from the kinematic correlation between @xmath2 and @xmath3 that plagued the caldwell plot . such preliminary hera data have recently became available @xcite@xcite . as we shall show , a pqcd analysis of these data is consistent with a sc interpretation @xcite . we follow the eikonal sc formalism presented in ref.@xcite , where screening is calculated in both the quark sector , to account for the percolation of a @xmath9 through the target , and the gluon sector , to account for the screening of @xmath6 . the factorizable result that we obtain is [ b ] = d_q(x , q^2)d_g(x , q^2 ) . sc in the quark section are given by [ c1 ] d_q(x , q^2)= db^2@xmath26 , [ c2 ] _ q = xg^dglap(x , q^2)(b^2 ) . the calculation is significantly simplified if we assume a gaussian parameterization for the two gluon non perturbative form factor , [ c3 ] ( b^2)=e^-b^2/r^2 . sc in the gluon sector are given by [ d1 ] xg^sc(x , q^2)=d_g(x , q^2 ) xg^dglap(x , q^2 ) , where [ d2 ] xg^sc(x , q^2)= _ x^1 _ 0^q^2dq^2 db^2@xmath27 . note that @xmath28 defined in . an obvious difficulty in the above calculation of @xmath29 stems from the fact that the @xmath30 integration spans not only the short ( pqcd ) , but also the long ( npqcd ) distances . to overcome this difficulty we assume that [ d3 ] xg@xmath31=xg@xmath32 , where @xmath33 . our choice of the above interpolation is motivated by the gauge invariance requirement that @xmath34 when @xmath35 . the sc calculation described above can be applied to any given input p.d.f . where the only adjusted parameters are @xmath36 and @xmath37 . as we shall see in the next section , @xmath38 is determined directly from the forward slope of @xmath5 photoproduction . @xmath37 is conveniently fixed at @xmath39 , the lowest @xmath2 value accessible for the input p.d.f . once we have chosen our p.d.f . , our sc calculation is essentially parameter free . we have checked that our output results are not sensitive to the fine tuning of these fixed parameters . our results are presented in figs . 1,2 and 3 . throughout this investigation we have used as input the @xmath40 version of grv98nlo . using the grv98dis version provides very similar results . following are some comments relating to our calculations and results : 1 ) as can be seen , in the limit of small @xmath2 and @xmath3 there is a significant difference between the screened and non screened values of @xmath23 . as expected the sc results are smaller and softer than the non screened input . 2 ) visibly , our overall reproduction of the experimental data is very good , in particular when considering that our input is essentially parameter free . a proper @xmath41 calculation requires the knowledge of the unknown theoretical errors . 3 ) if we follow the standard procedure and replace the theoretical errors with the experimental ones , we obtain excellent @xmath42 for 21 h1 data points with the exception of 3 points which are visibly out of line . the zeus data has errors which are considerably smaller and as a result our @xmath43 is not as good , even though our reproduction of the zeus data is reasonable . 4 ) the zeus @xmath44 and h1 @xmath45 data are somewhat softer than our predictions , which do not contain a soft non perturbative background . golec - biernat and wuesthoff @xcite have suggested studying the @xmath2 and @xmath3 behaviour of @xmath23 at fixed w as a method to determine the gluon saturation scales from the anticipated turn over in these plots . recent zeus low @xmath2 presentations @xcite of these plots show , indeed , the anticipated turn over structure in these figures , seemingly suggesting that gluon saturation is attained at @xmath46 . in our opinion the proper variables to study @xmath7 are @xmath3 and @xmath2 . trying to study the structure function by introducing other variables , such as w , may result in spurious effects which are predominantly kinematic . in the specific procedure suggested by golec - biernat and wuesthoff , the combination of the kinematic relation between @xmath47 and w with the very general behaviour of @xmath6 is sufficient to produce a turn over . its exact location depends on the details of the numerical input . consequently , the suggested fixed w plots do not convey any new dynamical information even if such information is hidden in the analyzed data . actually , it seems that any @xmath7 ( or @xmath48 ) parameterization which factorizes the @xmath2 and @xmath3 dependences , such as the buchmueller- haidt model @xcite , is capable of producing the fixed w turn over effects . to illustrate this point we consider the fixed w behaviour in two models which have very different dynamics : 1 ) our own glmn @xcite , which is a pure pqcd dipole model with sc . as such , our model relates indirectly to gluon saturation , even though it is constructed so as to include unitarity corrections below actual saturation . 2 ) the dl two pomeron parametrization @xcite , which is based on the regge formalism , and consists of the coherent sum of contributions from a `` hard '' and a `` soft '' pomeron , a normal reggeon and an additional contribution from the charmed sector which is proportional to the `` hard '' pomeron . each of these fixed j - poles are multiplied by a fitted @xmath49 form factor . the parameters of the `` model '' were determined from the requirement of a `` best fit '' to the experimental @xmath50 data , and are not directly associated with the gluon distribution in the proton or `` saturation '' . the results obtained for the logarithmic slope of @xmath51 at fixed w for both parametrizations are compared to the zeus experimental results in fig.4 and 6 . in figs.5 and 7 we also display the behaviour of @xmath52 at fixed @xmath49 and fixed x. we note that both `` models '' provide a reasonable description of the zeus data including the observed turn over in the fixed w plots.as grv98 is only applicable for @xmath53 , we have repeated the glmn calculation with grv94 which has @xmath54 and again reproduce the zeus fixed w turn over effect . from the above it appears that any model which provides a reasonable description of @xmath50 will exhibit a turn over in fixed w plots of the logarithmic derivative of @xmath50 , this occurs due to the relation between the kinematic variables x , @xmath49 and w , and is not related to `` saturation '' . however , the examples ( the buchmueller - haidt model and donnachie - landshoff approach ) which have been used to demonstrate that the saturation " is not a unique mechanism for a turn over at fixed @xmath55 , have one common feature : the soft " contributions are concentrated at a rather large typical scale @xmath56 . this observation supports the idea that the so called soft pomeron stems from rather short distances @xcite . therefore , an equally good description of the experimental data can be achieved either from gluon saturation " ( shadowing corrections ) or by matching soft and hard processes if the soft ones occur at rather short distances @xmath57 . in conclusion , whereas we find strong support for the need for sc in the small @xmath2 and @xmath3 limits for @xmath52 we are unable , as yet , to determine the gluon saturation scale directly from the latest hera data . the gluon saturation scale may be theoretically estimated from the contours produced at the boundary of @xmath58 , as discussed in our papers @xcite@xcite . the t=0 differential cross section of photo and dis production can be calculated in the dipole lla approximation @xcite@xcite@xcite@xcite@xcite . when using the static non relativistic approximation of the vector meson wave function @xcite@xcite , the differential cross section has a very simple form , viz . [ q ] @xmath59_0= @xmath60 ^ 2 @xmath61 , where in the non relativistic limit we have [ q1 ] \|q^2= and [ q2 ] x=. in the following we discuss in some detail the photo and dis production of @xmath5 . there is a lot of data available for this channel @xcite@xcite@xcite for the integrated cross section spanning a relatively wide energy range . from a theoretical point of view , its hardness ( or separation ) scale is comparable to the scales we have studied in our @xmath23 analysis . to relate the integrated cross section to we need to know @xmath62 , the @xmath5 forward differential cross section slope . for this we may use the experimental values , which are approximately constant . as we shall see , sc account well for the reported moderate energy dependence @xcite . the main problem with the theoretical analysis of @xmath5 is the observation that the simple pqcd calculation needs to be corrected for the following reasons : 1 ) a correction for the contribution of the real part of the production amplitude . this correction is well understood and is given by @xmath63 , where @xmath64 and @xmath65 . 2 ) a correction for the contribution of the skewed ( off diagonal ) gluon distributions @xcite . this correction is calculated to be r_g^2= @xmath66 ^ 2 . 3 ) a more controversial issue relates to the non relativistic approximation assumed for the @xmath5 charmonium . relativistic effects produced by the fermi motion of the bound quarks result in a considerable reduction of the calculated pqcd cross section @xcite . we denote this correction @xmath21 and note that it is very sensitive to the value of @xmath18 . ref.@xcite assumes that @xmath67 and obtains @xmath68 with minimal energy dependence . a small change in the input value of @xmath18 changes the above estimate significantly . we suggest , therefore , to consider @xmath21 as a free parameter . in our calculations we have used @xmath69 which corresponds to a c - quark mass of approximately @xmath70 . since the @xmath71 photo and dis cross sections are proportional to @xmath13 , the study of this channel can serve as a compatibility check supplementing our study of @xmath23 . as noted in the introduction our sc approach was triggered by the observation that none of the latest p.d.f.s can provide a good simultaneous reproduction of the two channels under consideration . our calculation of sc for @xmath5 photo and dis production is rather similar to the @xmath2 logarithmic slope calculation presented in the previous section . we follow our earlier publication @xcite and define the damping factors due to the screening in the quark sector i.e. the percolation of the @xmath72 through the target . this is given by the following expressions for the longitudinal and transverse dampings [ scl ] d_ql^2= and [ sct ] d_qt^2= our expression for @xmath73 , the damping in the gluon sector is the square of the gluon damping defined in the previous section . our final expression for the forward cross section is [ final ] @xmath74_0= c_r^2 c_g^2 c_f^2 @xmath75_0^pqcd d_q^2 d_g^2 , where @xmath76 denotes the l and t components as appropriate . our calculations as compared with the data are presented in figs.8 and 9 . as can be seen , our reproduction of the data is excellent with a @xmath77 which is well below 1 . these excellent @xmath41 values are maintained when calculating over the entire data base as well as limiting ourselves to the high energy hera data . note that @xmath8 , which is the essential parameter in the sc calculation is determined directly from the @xmath5 photoproduction forward slope . in a model such as ours , we expect a weak dependence of @xmath78 on the energy @xcite ( see fig.10 ) . we have shown that nlo grv98 when screened both for the percolation of a @xmath79 pair through the target , and for multigluon exchange provides an excellent repoduction of both the h1 and zeus data for the logarithmic derivative of @xmath50 structure function of the proton . the experimental data and our model are consistent with @xmath80 ( at fixed @xmath49 ) being a monotonic increasing function of 1/x . no deviation of this behaviour has been seen even at the lowest values of @xmath49 and x. the suggested turn over seen in @xmath81 at fixed w , as a function of x or @xmath49 , does not appear to be connected with saturation effects , and can not be used as a discriminator , as it appears in all parametrizations of @xmath51 which provide a reasonable description of the data . the same screened nlo grv98 also provides an excellent description of the photo and dis production of @xmath82 , once corrections are made for the real part of the production amplitude , for the skewed ( off diagonal ) gluon distribution and for the fermi motion of the bound quarks in the charmonium wave function . in a seperate publication @xcite we show that we are unable to obtain a simultaneous good description of @xmath52 and the photo and dis production of @xmath83 using other pdf s on the market e.g. mrs99 , cteq4 and cteq5 even after including screening corrections . only screened nlo grv98 is able to provide a satisfactory simultaneous description of the latest hera data available for @xmath84 and photo and dis production of @xmath82 . 99 e. gotsman , e . ferreira , e. levin , u. maor and e. naftali , _ ( to be published ) _ . gribov , e.m . levin and m.g . ryskin , ; a.h . mueller and j. qiu , ; l. mclerran and r. venugopalan , , , , . e. levin and m.g . ryskin , ; j.c.collins and j. kwiecinski , ; j. bartels , j. blumlein and g. shuler , ; e. laenen and e. levin , and references therein ; a.l . ayala , m.b . gay ducati and e.m . levin , , ; yu . kovchegov , , , , _ phys . rev . _ * d61 * ( 2000)074018 ; a.h . mueller , _ nucl . phys . _ * b572(2000)227 , ; yu . v. kovchegov , a.h . mueller , ; i. balitsky ; e. levin and k. tuchin , _ nucl . b573(2000 ) 833 ; j. jalilian - marian , a. kovner , l. mclerran and h. weigert , ; j. jalilian - marian , a. kovner and h. weigert , ; j. jalilian - marian , a. kovner and h. weigert , ; j. jalilian - marian , a. kovner , a. leonidov and h. weigert , , erratum - ibid . ; a. kovner , j.guilherme milhano and h. weigert , outp-00 - 10p , nordita-2000 - 14-he , hep - ph/0004014 ; h. weigert , nordita-2000 - 34-he , hep - ph/0004044 ; m. braun lu - tp-00 - 06,hep - ph/0001268 . e. gotsman , e.m . levin and u. maor:;. m. gluck , e. reya and a. vogt : _ eur . phys . j. _ c5 * ( 1998 ) 461 . martin , r.g . roberts , w.j . stirling and r.s . thorne , _ eur . phys . j. _ * c4 * ( 1998 ) 463 . cteq collaboration , h.l . lai et al . : hep - ph/9903282 . m. klein ( h1 collaboration),talk given at 19-th int . symposium on lepton - photon interaction at high energy , hep - ex/0001059 . b. foster ( zeus collaboration ) : invited talk , royal society meeting , london , may 2000 ; zeus collaboration , j. breitweg et al . : desy 00 - 071 . zeus collaboration , j. breitweg et al . : desy 99 - 160 and quoted earlier zeus publications . h1 collaboration , c. adloff et al . : hep - ex/0003020 and quoted earlier h1 publications . e401 collaboration , m. binkly et al . : ; e516 collaboration , b.h . denby et al . : ; e. gotsman , e.m . levin and u. maor : . e. gotsman , e.m . levin , u. maor and e. naftali : . k. golec - biernat and m. wuesthoff : . shuvaev , k.j . golec - biernat , a.d . martin and m.g . ryskin : hep - ph/9902410 ; a. freund and v. guzey : . l. frankfurt , w. koepf and m. strikman:;. a.l . ayala filho , m.b . gay ducati and e.m . a. caldwell ( zeus coll . ) : invited talk , desy theory workshop , hamburg , october 1997 ; zeus collaboration , j. breitweg et al . : . m. gluck , e. reya and a. vogt : _ z. phys . _ * c67 * ( 1995 ) 433 . kaidalov , c. merino and d. pertermann : hep - ph/0004237 ; a. capella , e.g. ferreiro , c.a . salgado and a.b . kaidalov : hep - ph/0005049 ; hep - ph/0006233 . a. donnachie and p.v . landshoff : ; . h. abramowicz , e.m . levin , a. levy and u. maor : ; h. abramowicz and a. levy : desy 97 - 251 and hep - ph/9712415 . w. buchmueller and d. haidt : desy-96 - 061 and hep - ph/9605428 . d. haidt : _ nuc . phys . suppl . ) _ * b79 * ( 1999 ) 186 . thorne at low x workshop , oxford , july 2000 ) . d. kharzeev and e. levin , _ nucl . * b578 * ( 2000 ) 351 . m.g . s.j . brodsky et al . : . ryskin , r.g . roberts , a.d . martin and e.m . l. frankfurt , m. mcdermott and m. strikman : hep - ph/9812316 . martin , m.g . ryskin and t. teubner : hep - ph/9901420 . e. gotsman , e.m . levin and u. maor:;. e. gotsman , e.m . levin and u. maor : .	we expect that s - channel unitarity should materialize in hard dis reactions through screening corrections ( sc ) indicating that the gluon distribution function is approaching saturation , it is not as yet clear what the kinematical scales are at which these effects become important . while the global dis @xmath0 total cross section , or @xmath1 , data are well reproduced by dglap evolution without substantial sc , there exists experimental data from hera which suggests deviations from dglap predictions in the small @xmath2 and @xmath3 limits . these signatures are observed in both the fine details of @xmath1 provided @xmath2 and @xmath3 are small enough , as well as in the diffractive channels . in this investigation we present a detailed study of @xmath4 which is supported by a coupled analysis of @xmath5 photo and dis production . both channels are directly proportional to @xmath6 , and as such serve as excellent discriminators between different approaches and models . in the first phase of our investigation we have found that none of the latest editions of the parton distribution functions ( grv98 , mrs99 , cteq5 ) provides an adequate and simultaneous reproduction of @xmath2 logarithmic slope of @xmath7 at small @xmath2 values as well as @xmath5 photoproduction ( details of this will be published separately @xcite ) . we then show that taking grv98nlo as input and correcting it for sc , we can reproduce the recent hera data well . the calculation depends on one parameter @xmath8 which is directly determined from the @xmath5 photoproduction forward differential slope . with this input we obtain an excellent fit to the @xmath5 photo and dis production data . our calculations made in the lla of pqcd take into account the corrections implied by the real part of the production amplitude , off diagonal ( skewed ) gluon distributions and the fermi motion of the charm quarks within the bound charmonium system . the sc are consistently calculated for both the percolation of a @xmath9 through the target and the screening of the gluon parton distribution which forms the base of our calculation . our main conclusion is that , whereas we find strong support for the need for sc in the small @xmath2 and @xmath3 limits of the channels we have investigated , the latest hera data is not sufficiently precise to directly determine the gluon saturation scale . + * paper no.489 * + * july 2000 *
low mass x - ray binaries are for the majority of the time in a state of quiescensce with faint or non - detected x - ray emission , though optical or near - infrared ( nir ) counterparts may be visible due to emission from the donor star , or possibly the jet , hot spot , or outer accretion disk . they are often only discovered when they enter an active state of outburst when powered by an increased level of accretion onto the central , compact object ( black hole or neutron star ) there is a dramatic increase of the x - ray , optical / nir and radio flux . during these outbursts the systems have been observed to go through a number of high energy spectral states before returning to a quiescent state , usually on times scales of weeks , months or even longer . these states are a generally low intensity , power - law dominated , _ hard _ state followed by a usually , higher intensity , _ thermal - dominant _ , _ soft _ state which decreases in flux , via a late hard state , over time . additionally , the hard states are associated with aperiodic variability of the light curve not present in the soft state ( see @xcite for a fuller description of the various possible states ) . xtej1752 - 223 , a new x - ray transient and black hole candidate @xcite in the galactic center region , was detected on 2009 - 10 - 23 at 19:55 ut ( mjd 55128.33 ) by rxte and on 2009 - 10 - 24 at 14:18:50 ut by the bat instrument on board the _ swift _ satellite @xcite . the high energy , variable emission of the source was confirmed in the following days by _ swift_/xrt @xcite and rxte @xcite as well as by maxi / gsc @xcite and _ fermi_/gbm @xcite . an optical and nir counterpart was proposed by @xcite while a radio source coincident with the x - ray position @xcite was observed by the australia telescope compact array ( atca ; @xcite ) . the source was observed to have undergone a state transition in mid - january 2010 ( mjd @xmath0 ; @xcite ) , from a spectrally hard to a spectrally soft x - ray state . this was identified by a de - coupling of high energy ( @xmath1kev ) and low energy ( @xmath2kev ) light curves @xcite which had previously traced each other , and by the dominance of a thermal component . the source was observed to have reverted to a hard state at the end of march 2010 ( mjd @xmath3 ; @xcite ) , after the low energy light curve decreased to trace once more the high energy light curve ; the thermal component was no longer dominant . it has also been suggested that the turn - on of a compact jet has been observed in this system @xcite during this transition back to the low / hard state . in this article we present the _ swift _ burst alert telescope , x - ray telescope and ultraviolet / optical telescope monitoring observations of xtej1752 - 223 , obtained over the period of the outburst . based on these data , we identify the periods of the various states and compare the behavior of the major photometric , spectral and timing parameters during these states to those expected from black hole x - ray binaries . in section [ section : observations ] we introduce the observations and reduction methods , while in section [ section : analysis ] we present the results of our photometric , spectral and timing analyses of the data . we summarise our findings in section [ section : conclusion ] . _ swift _ @xcite observed xtej1752 - 223 with its narrow field instruments , the ultraviolet / optical telescope ( uvot ; @xcite ) and the x - ray telescope ( xrt ; @xcite ) , on 55 epochs from october 25 to november 03 2009 and from february 04 2010 to june 5 2010 . these observations ( under target ids 31532 , 31640 and 31688 ) coincide with the period of the outburst . observations with the narrow field instruments were interrupted for 3 months due to the source becoming constrained by the position of the sun . wide field instruments also became sun constrained but at a later date and for a shorter period of time , depending on the satellite and instrument in question ( e.g. maxi was constrained from december 15 to 24 ) . during the period of the outburst , as it had been on previous occasions , the field was also being observed ( though constrained from december 13 to 28 ) by _ swift _ s burst alert telescope ( bat ; @xcite ) which , in combination with xrt and uvot , affords quasi - simultaneous observations from 0.002 to 150 kev . reduction of _ swift _ data was carried out using version 3.4 of the _ swift _ software , released as part of heasoft 6.7 . all uncertainties are quoted at the @xmath4 confidence level . uvot data have been pre - processed at the _ swift _ data center ( see @xcite ) and require only minimum user processing . as the pointing of _ swift _ is only accurate to a few arcseconds , the standard pre - process pipeline provides an astrometric correction by matching sources with the usno - b catalog . however , in this case the automatic matching failed because the field differs significantly between the ultraviolet and the usno - b bandpass . to overcome this , we created our own astrometric reference catalog by selecting blue and bright field stars in the gsc-2.3 catalog @xcite to determine astrometric corrections for the data . the image data of each filter , from each observation sequence , i.e. , with a given observation i d , were summed using uvotimsum . all sequences were then combined in an individual fits file , using fappend , which was again summed using uvotimsum to obtain a deep image of the field in each filter . from these images , two possible optical counterparts of xtej1752 - 223 are identified , consistent with the x - ray position @xcite ; source a , consistent with the counterpart proposed by @xcite , is within @xmath4 and source b is within @xmath5 of the x - ray position . the positions ( j2000 ; 0.31error ) of the possible counterparts were derived from such a deep ( 1853s ) @xmath6 band image summed between mjd 55130 and 55138 ( figure [ fig : uvot_v ] ) , using the uvotdetect command , as : + [ cols= " < , < , < , < " , ] the xrt windowed timing mode ( wt ; @xcite ) event data [ 0.3 - 10kev ] were analysed using the tools presented by @xcite . these correct for pile up and the affect of bad ccd columns . version v011 response matrices were used for the spectral analysis . the best xrt position of 17:52:15.14 @xmath722:20:33.8 ( j2000 ; 3.1error ) is as previously reported @xcite . background subtracted xrt count rate light curves are binned per observing sequence ( figure[fig : xray_lc ] ; combined with maxi observations ) . unbinned source light curves ( not background subtracted ) , for each sequence , were used to produce power density spectra ( pds ) and to calculate the rms variability with the ftools , powspec and lcstats respectively . within each of these the binning was set to be n times the minimum bin size ( 1.766ms ) where n=1 for the pds and n=8 ( 14.128ms ) for the rms variability , giving frequency ranges of 0.02 - 570hz and 0.02 - 70hz respectively . the average pds were calculated , within powspec , from an input list of the light curves of the required sequences , using the same binning . lags between the standard xrt low [ 0.3 - 1.5kev ] and high [ 1.5 - 10kev ] energy band unbinned light curves were calculated with crosscor and a bin size set to n=8 ( 14.128ms ) . bat detector plane histograms ( dphs ) were obtained in survey mode ; these are 2 dimensional images with full spectral data and coarse time binning . spectra and light curves were extracted from these dphs using mask weighting ( also known as ray - tracing ) to account for the position of the source on the coded aperture mask . the data were initially processed with the ftool , batsurvey which applies data quality cuts and uses the tools batfftimage and batcelldetect to extract 8-channel spectra for the source . we then used custom software to format the spectra for xspec , and batdrmgen to generate the response files . the light curve was derived using a similar method . bat , as a survey telescope , had been observing the field of xtej1752 - 223 since the start of _ swift _ operations in january 2006 , though not monitoring the source . the extracted bat light curve ( 15 - 150kev ) of the source from january 12 2006 through to the present , displays no activity until the 2009 outburst . the joint xrt - bat spectra were fit within xspec ( 12.5.1 ) , using c - statistics and interstellar absorption modeled by tbabs ; initially we fit an absorbed power law ( pow ) to the spectra before adding a thermal component ( diskbb ) . we note that these are phenomenological models for the respective spectral states , not a suggestion that these are necessarily the best physical models ; they should not be used to deduce physical constraints but only as an indication of the spectral state of the source . the early data ( mjd 55131 - 55138 ) are well described by the power law of average photon index , @xmath8 , ( figure[fig : spectral ] ) and do not require a thermal component - clearly indicating that the source is in the hard state . the absorption of a simultaneous fit of these sequences is @xmath9@xmath10 , greater than the galactic value of @xmath11@xmath10 but lower than the @xmath12@xmath10 found by @xcite . data observed from mjd 55234 - 55280 additionally support a soft state thermal component decreasing from a temperature of @xmath13kev to @xmath14kev over the period , as well as decreasing normalisation . due to the degeneracy between the power law and the thermal components and lack of counts from the bat , the power law index is poorly constrained in a manner not obvious from the nominal fit errors but causing a high level of variability in the upper panel of figure[fig : spectral ] ; these values are not reliable . from mjd 55283 the data no longer support a thermal component , though the signal to noise at these low counts rates can not rule out a thermal component . additionally , as can be seen from figure[fig : spectral ] , the power law indices are decreasing until mjd 55328 , signaling the transitional period between the soft and hard states , in agreement with the increasing hardness ratio ( figure[fig : xray_lc ] ) . from mjd 55328 to the end of our observations , on mjd 55352 , the spectra are described by a power law of relatively constant photon index , the average of which is @xmath15 . the power law indices and the hardness ratios at these late times fail to return to the original , rising hard state values ( @xmath8 ) , indicating that , though the count rate is lower than it was in that rising hard state , the source has not yet returned to its original hard state or that the late hard state has marginally different spectral properties . an optical and nir counterpart was proposed by @xcite based on a comparison between their images and those from the digitized sky survey , 2mass and ukidss . we have confirmed uvot source a as this optical counterpart on the basis of a significant , though low amplitude variability in the @xmath6 , @xmath16 and @xmath17 bands ( figure[fig : uvot_lc ] ) . furthermore , these optical magnitudes , or count rates , exhibit a power - law correlation with the x - ray count rate , as observed by xrt at times after mjd 55232 ( figure[fig : uvot - xrt ] ) , i.e. , when the source is in the soft state and later , during the transition to the late hard state . the earlier , hard state data does not fit on this correlation , but displays hysteretical behaviour and for the given x - ray count rate it spans has a significantly higher magnitude in all bands . this is similar to the hysteretical behaviour observed in the nir for a number of transients @xcite , most notably xtej1550 - 564 , where the additional hard state emission is attributed to optically thin synchrotron emission from a jet @xcite , which would be quenched in the soft state ( e.g. , @xcite ) , and weak at low x - ray luminosities , leading to the hysteresis . though this hysteresis is not normally observed in the optical bands , the optical data presented here is in good agreement with the synchrotron emitting jet making a significant contribution to the optical emission in the rising hard state . while the observed @xmath6 band brightening from mjd 55292 corresponds to the suggested turn - on , in nir , of a compact jet between mjd 55293 and 55298 @xcite , it is not corroborated by the other bands and is not statistically significant , though it may indicate the turn - on of a jet and the transition to the hard state ; at such low x - ray luminosity levels , the jet contribution to the optical would be minimal . from a simultaneous fit of the @xmath6 , @xmath16 and @xmath17 bands the correlation is found to be @xmath18 with observed colours of @xmath19 and @xmath20 ( @xmath21 ) , though the individual bands display indices of @xmath22 , @xmath23 and @xmath24 respectively , hinting at possible spectral evolution . such a correlation is more usually associated with the hard state @xcite but we can not say if the early , hard state data follows another correlation ( expected to be @xmath25 ; @xcite ) due to the very limited range of magnitudes and x - ray count rates over this period . though we find a correlation between the optical and x - ray , we can not imply causality or say which leads without a study of the phase lag between the two bands , which is not possible given the time resolution of the optical data . similarly a timing analysis to investigate variability of the optical data is not possible . the correlation may be due to reprocessing of the x - rays or through both originating from the accretion disk ; it is also possible that optical emission in the soft state originates from a jet , the state of which traces the spectral states @xcite the quality of the uvot spectral data does not support any spectral evolution in the optical , nor does it enable us to specify the most appropriate spectral model for the source such as a power law associated with the above suggested jet emission . hence , we model the data phenomenologically in xspec , by a power law ( @xmath26 ) and a dust extinction component ( zdust ) , though we note that this is not a particularly good fit . given that an unknown amount of galactic extinction , @xmath27 @xcite , effects the data we can only place limits on the spectral index of @xmath28 , though the nominal best fit is at @xmath29 for @xmath30 . given the uncertainties of the spectral slope and the lack of phase lag information , we are unable to make a meaningful comparison to the predictions of the various models . the hardness - intensity diagrams ( hid ) for both the xrt data [ 1.5 - 10kev / 0.3 - 1.5kev ] and the quasi - simultaneous ( same day ) bat / maxi data starting at the same time are shown in figure[fig : hid ] . due to the fact that the bat / maxi data are only quasi - simultaneous , the hardness ratio ( hr ) of bat [ 15 - 150kev ] with respect to maxi [ 4 - 10kev ] at low bat count rates ( @xmath31 ) , where the relative variability over a day is greater , are underestimated by the error bars . in general , hids follow a canonical trajectory ( e.g. , @xcite ) : increasing intensity at a high hr to a maximum in the hard state , indicating the power law shape of the spectra ; decrease of hr at a high intensity through an intermediate state as the thermal component of the spectra becomes more dominant ; decrease of intensity at the low hr level during the soft state ; before the hr increases at low intensity , through an intermediate state , to the original value . this is consistent with both the xrt data , though a significant amount of the trajectory in the first intermediate state is missing due to the source being sun constrained for a period of 3 months , and the bat / maxi data in which the final phase is obscured due to the large error bars on the hr . it is clear from the xrt hid that the hardness has yet to increase to the original value , indicating that the source has not yet returned to its original hard state , though the count rate is lower than it was in that rising hard state , or that the late hard state has different properties . these tracks are common among black hole binary sources though are also observed in neutron star binary sources such as aquilax-1 ( e.g. , @xcite ) and further confirm the spectral state of the system over the periods in question . the rms variation of the light curves exhibit a similar evolution to hardness ratio ( figure[fig : xray_lc ] ) , from average rms @xmath32 per cent in the observed period of the rising hard state ( mjd 55131 - 55138 ) to @xmath33 percent during the soft state observations and increasing to the original levels at later times . examining the light curves in the standard xrt low [ 0.3 - 1.5kev ] and high [ 1.5 - 10kev ] energy bands we find that , during the rising hard state this variability is more pronounced in the low band with an average ratio between the two of @xmath34 . this suggests an energy dependency of rms variation , in contrast to @xcite who find a flat rms spectrum @xcite from xte data at energies in the range 2 - 10kev . the lag between the two bands , calculated via a cross correlation , is consistent with zero in all cases though is not tightly constrained . examining the average power density spectra of the states ( figure[fig : powspec ] ) we see that the hard state has a high level of power in low frequency ( @xmath35hz ) variability , though there is no significant sign of quasi - periodic oscillations ( qpos ) . the soft state , on the other hand , has a much lower lever of power and only at frequencies @xmath36hz , as is manifested by the @xmath33 percent limit of rmss during this period ( mjd 55234 - 55289 ) . after mjd 55321 , where the xrt count rate drops below 10 cts / s , the light curves and pdss are heavily affected by background counts ( figure[fig : powspec_back ] ) and hence the rms values for these observations are unreliable and not plotted in figure[fig : xray_lc ] . the excess of low energy variability in the rising hard state is similar to the behaviour of the hard state variability in swiftj1753.5 - 0127 and gx339 - 4 as reported by @xcite . within the framework of their model , enhanced low energy variability ( at frequencies @xmath37hz ) is due to a contribution from the thermal emission of the accretion disk . the excess variability is attributed to intrinsic disk variability due to accretion rate fluctuations as opposed to reprocessing of the spectral power - law component of the emission . though a thermal component is not supported by the spectra of xtej1752 - 223 in the rising hard state , the excess variability we observe suggests that one might contribute to the flux at low energies . more detailed spectral fits using the methods introduced in @xcite are required to test this possibility . data , over a greater energy range , which better constrains column density and photon index may also be necessary . xtej1752 - 223 was observed by _ swift _ in a relatively low intensity , hard state from 2009 - 10 - 26 to 2009 - 11 - 03 ( mjd 55131 - 55138 ) which was signified by hard power law energy spectra , and high levels ( @xmath38 per cent ) of rms variability in the x - ray light curve . xtej1752 - 223 then became constrained by the position of the sun and was unobservable until 2010 - 02 - 04 ( mjd 55234 ) by which time it had experienced a ten fold increase in the flux observed by xrt [ 0.3 - 10.0kev ] , which proceeded to decrease to the original levels over a period of 2 months . the spectra during this period clearly exhibited a strong thermal component and the variability and hardness ratios of the light curves had dropped significantly , consistent with the source being in a soft state until 2010 - 03 - 25 ( mjd 55280 ) . observations from 2010 - 03 - 28 ( mjd 55283 ) to 2010 - 05 - 12 ( mjd 55328 ) do not support a thermal component but do display a decreasing photon index along with increasing hardness and rms variability . this suggests a transitional state before reverting to the hard state , with a relatively constant photon index , in which the system is observed until the end of observations on 2010 - 06 - 05 ( mjd 55352 ) . the high energy hardness - intensity diagrams over two separate bands follow the canonical behavior associated with a black hole binary , further confirming the spectral states of the system during the outburst . our timing analysis shows that in the hard state there is significant variability below 10hz which is more pronounced at low energies , while during the soft state the level of variability is consistent with being minimal . we are able to confirm that the uvot source at the position of the proposed optical and near - infrared counterpart is associated with xtej1752 - 223 due to the observed low amplitude variability and , in the soft state and later , correlation with the x - ray emission as measured by xrt . however , we can not state that the optical flux is causally connected to the x - ray emission via reprocessing of the x - rays or through both originating from the accretion disk ; given the uncertainties of the spectral slope and the lack of phase lag information , we are unable to make a meaningful comparison to the predictions of the various models . the optical counterpart also displays hysteretical behaviour not normally observed in the optical bands ; for a given x - ray count rate , the magnitude in the rising hard state is significantly higher than that in the soft state . this is similar to the hysteretical behaviour observed in the nir for a number of transients where the additional hard state emission is attributed to optically thin synchrotron emission from a jet , which would be quenched in the soft state and weak at low x - ray luminosities . though this hysteresis is not normally observed in the optical bands , it is in good agreement with the synchrotron emitting jet making a significant contribution to the optical emission in the rising hard state . we thank the referee for constructive comments and paul kuin for useful discussions on uvot . pac , pae and cb acknowledge support from stfc . tjm thanks the eu fp7 for support through grant number itn 215212 ` black hole universe ' . pc acknowledges funding via a eu marie curie intra - european fellowship under contract no . 2009 - 237722 . this research has made use of : swift data supplied by the uk swift science data centre at the university of leicester ; and maxi data provided by riken , jaxa and maxi teams . , j. e. , et al . 2004 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 5165 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . k. a. flanagan & o. h. w. siegmund , 217	we present _ swift _ broadband observations of the recently discovered black hole candidate , x - ray transient , xtej1752 - 223 , obtained over the period of outburst from october 2009 to june 2010 . from _ swift_-uvot data we confirm the presence of an optical counterpart which displays variability correlated , in the soft state , to the x - ray emission observed by _ swift_-xrt . the optical counterpart also displays hysteretical behaviour between the states not normally observed in the optical bands , suggesting a possible contribution from a synchrotron emitting jet to the optical emission in the rising hard state . we offer a purely phenomenological treatment of the spectra as an indication of the canonical spectral state of the source during different periods of the outburst . we find that the high energy hardness - intensity diagrams over two separate bands follows the canonical behavior , confirming the spectral states . our xrt timing analysis shows that in the hard state there is significant variability below 10hz which is more pronounced at low energies , while during the soft state the level of variability is consistent with being minimal . these properties of xtej1752 - 223 support its candidacy as a black hole in the galactic centre region . [ firstpage ] x - rays : binaries x - rays : bursts binaries : close stars : individual : xtej1752 - 223
the fine structure lines of neutral atomic carbon , ci , at 809.3435 ghz and at 492.1607 ghz ( cooksy et al . @xcite ; yamamoto & saito @xcite ) provide an important diagnostic tool for the physical and chemical conditions of the dense interstellar medium and contribute significantly to the energy balance of the gas . after the first detections (: phillips et al . @xcite ; : jaffe et al . @xcite ) the rapid advance in receiver sensitivity , driven by the technological progress of superconducting mixers now reaching sensitivities of only a few times the quantum limit even at submillimeter wavelengths , has resulted in the line now being well studied in many galactic sources . the three detections of the line in external galaxies ( ic342 : bttgenbach et al . @xcite ; m82 : schilke et al . @xcite , white et al . @xcite ; ngc253 : harrison et al . @xcite ) show that ci exhibits rather different properties in these starburst galaxies from the ones in the milky way as measured by cobe ( wright et al . @xcite ; bennet et al . @xcite ) . both in m82 and in ngc253 the ci / co abundance ratio is rather high , 0.4 to 0.5 , whereas it is only about 0.1 even in massive star forming regions in the milky way . observations of the upper , , transition are still rather rare as only few receiver systems are operational at these wavelengths . the fine structure line intensity ratio is of particular astrophysical importance . in the optically thin regime the integrated line intensity is proportional to the upper state column density . the line intensitiy ratio thus directly measures the excitation temperature , @xmath5 . measuring the integrated line intensity on a scale , @xmath6 , and using the values for the a - coefficients given by nussbaumer & rusca ( @xcite ) , we obtain @xmath7 $ ] . more generally , an escape probability radiative transfer model , covering also the case of higher optical depth , links the observed line ratio with the physical parameters of the source . in this _ letter _ we report the first extragalactic detection of the line , which was observed from the starburst galaxy m82 simultaneouly with the line , thus providing a very good relative calibration of the line ratio and hence allowing an accurate estimate of the excitation conditions in the emitting gas . the two neutral atomic carbon fine structure lines were observed simultaneously using a new dual channel receiver built at the universitt zu kln ( honingh et al . both channels have a tunerless sis waveguide mixer furnished with sis junctions designed and fabricated in the cologne microstructure laboratory . one operates from 780 to 820 ghz in dsb mode with @xmath8 ; the other , from 460 - 490 ghz , is ssb tuned with a martin - puplett interferometer . due to technical problems with a cooled if - amplifier , the low frequency channel had a @xmath9 of only @xmath10 , rather poor in comparison with its standard performance ( @xmath11 ) , but still better than the high frequency channel and thus not limiting the sensitivity for the simultaneous measurement . these observations , performed on march 3 , 1996 , are the first heterodyne measurements at frequencies above 460 ghz with the new 10 m heinrich hertz telescope ( baars et al . @xcite ) on mt . graham , arizona , usa . the receiver package includes a dual channel acousto - optical spectrometer backend with a 1.1 ghz bandwidth each and 1mhz resolution ( schieder et al . calibration is done via an internal hot load / cold load / sky - flip mirror , following the procedure outlined in harris ( @xcite ) and including corrections for atmospheric sideband imbalance derived from standard atmospheric models ( grossman @xcite ) . the zenith transmission was 0.41 at 809 ghz and 0.54 at 492 ghz for an 8 hour period during which the m82 data were taken . note that the ratio of transmissions predicted by the standard atmospheric model is 0.96 . the difference is likely caused by a vertical structure of the atmosphere different from the model assumptions . line intensities are given in units of @xmath12 . main beam efficiencies were determined according to @xmath13 , where @xmath14 is the antenna temperature measured on a planet , @xmath15 the rayleigh - jeans temperature of the planet disk , and @xmath16 $ ] the geometric coupling factor between the planet ( disk diameter @xmath17 ) and the gaussian beam . @xmath18 was derived for the 9 `` fwhm 809 ghz beam , and @xmath19 for the 15 '' fwhm 492 ghz beam from measurements of venus ( 17 " diameter ) , assuming its physical temperature to be 280k ( altenhoff @xcite ) . subsequent realignement of the secondary mirror improved these efficiencies , but is irrelevant for the data presented here . due to the simultaneous observation of both lines , the line intensity ratio is measured very precisely . combining the above relations , the line ratio is effectively calibrated against the observed intensity ratio of the planet and is given by @xmath20 where @xmath21 is the difference of the atmospheric opacity at both line frequencies . indices l and p distinguish the _ line _ and _ planetary _ measurements . as the atmospheric transmission at the fine structure transition frequencies is of comparable magnitude , and as long as the planetary calibration measurement and the source observations are done at roughly the same transmission , the @xmath22 $ ] factor is always close to unity and depends only weakly on the actual transmission . the error in the line ratio is thus dominated by the uncertainty in the coupling factor , i.e. the uncertainty of the beam sizes at the two frequencies . we did not have a smaller planet than venus available for calibration at the period of the observations , and were thus not able to measure the beam sizes directly with sufficient precision . instead , we use the nominal sizes quoted above , corresponding to the diffraction limit of a cassegrain optics with 14 db edge taper ( goldsmith @xcite ) . the observations were done using the secondary chopping mirror with a 2 arcmin throw at 2 hz in azimuth and alternatingly pointing the telescope with the source in one and the other chopper beam . as the observations were done in one of the first nights after the receiver was installed , we did not yet have a good pointing model available . we nevertheless found emission from the sw lobe of m82 , clearly identifiable via its line shape and velocity . we mapped several positions on a 5 " grid around the position of strongest emission . unfortunately , we were not able to extend our observations to the center and the ne lobe of m82 due to technical problems with the telescope drive software after about two hours of observing . pointing observations over the subsequent days confirmed the m82 observations to be pointed on the sw lobe of the galaxy . figure 1a , b ) shows the positionally averaged spectra obtained by summing 33 scans of 2 min duration each , pointed at 8 positions within a 7 `` radius from the position of maximum signal . it thus represents the emission in a beam of effectively 14 '' fwhm for the 809 ghz and 20 `` for the 492 ghz line . both lines are clearly detected with their line shape consistent with the earlier observations of the 492 ghz emission from the sw lobe by schilke et al . ( @xcite ) . the intensity of the 492 ghz line agrees with that reported in the same beam size by those authors . no baseline correction has been applied to the spectra displayed in panel a ) and b ) . the observed continuum offsets ( 0.6 k ( dsb ) at 809 ghz , i.e. 23 jy in the effective 14 '' beam ; 0.15 k ( ssb ) at 492 ghz , i.e. 8 jy in the effective 20 " beam ) are within the errors consistent with the broadband dust continuum fluxes ( 450 @xmath23 m , smith et al . @xcite ; 800 @xmath23 m , hughes , gear & robson @xcite ; 1300 @xmath23 m , krgel et al . @xcite ) . at 809 ghz the intensity at the map center is significantly higher than the map average ( fig . the line emission is thus more compact than the map extent . the 492 ghz line does not show as big an effect ( fig . 1b ) , consistent with it being observed with the larger beam of 15 " fwhm anyway . for the analysis of the line ratio we thus compare the 809 ghz map average with the 492 ghz center spectrum ( fig . the integrated intensity is 152 k for the and 158 k for the line , @xmath24 , which , as discussed above , should be accurate to within 10% . the intensity ratio in m82 is thus much higher than in the milky way : the large scale cobe maps ( bennett et al . @xcite ) give @xmath25 for the galactic center emission , and @xmath26 for the inner galactic disk . following the derivation in the introduction , the excitation temperature of the transition , and hence the minimum kinetic temperature of the emitting gas in m82 , is @xmath27 . the corresponding value from cobe for the galactic center is @xmath28 , for the inner galactic disk it is @xmath29 . fig.2 shows the result of an escape probability radiative transfer calculation ( stutzki & winnewisser @xcite ) for spherical clumps , using the collisional rate coefficients by schrder et al . ( @xcite ) . it confirms the minimum temperature of 50k . at this temperature the line ratio implies a gas density of at least @xmath30 . at higher temperatures a lower density is possible ( @xmath31 at @xmath32=120k ) . densities higher than the quoted minimum require the line to be optically thick and thermalized , and hence can only be reached at very high ci - column densities ( on the horizontal branch of the according contour lines in fig.2 ) . the minimum kinetic temperature and density derived are in good agreement with the estimates for the excitation conditions of the bulk co emission from m82 ( turner et al . @xcite , harris et al.@xcite , wild et al . @xcite , gsten et al . the -line ratio constrains the pressure of the interstellar medium in m82 to be @xmath33 k or higher . the absolute line intensities correspond to a beam averaged ci column density of @xmath34 . this value depends only weakly on the temperature and density of the gas ( see fig.2 ) . this is plausible , as the partition function of the three level system ci does not change significantly above a @xmath35 of about 20k , as was noted already by schilke et al . comparison with the total co column density of @xmath36 derived by wild et al . ( @xcite ) both from the low - j intensities and an excitation model for all observed co lines , then gives a high ci / co abundance ratio of 0.5 , confirming the high value found by schilke et al . ( @xcite ) and white et al . ( @xcite ) . the reason for the high ci / co ratio in m82 in comparison to typical galactic sources with a similar uv field intensity and comparable densities , is unclear . it might primarily be a geometrical effect , where a much smaller fraction of the presumably highly clumped material is located at sufficient column density to be well shielded from the uv radiation than is the case for typical galactic sources ; the cause for this would , of course , be linked to the rather different physical environment created by the starburst activity in m82 . schilke et al . @xcite argue that at the enhanced cosmic ray flux in m82 ( suchkov et al . @xcite ) , resulting from the high sn rate ( kronberg et al . @xcite ) following the starburst , the high ionisation solution found by lebourlot et al . ( @xcite ) for low density material in dark interstellar clouds with its correspondingly large ci / co ratio will be a stable solution of the chemical network even at the high densities of the interstellar medium in m82 . the high cosmic ray flux also explains the high temperature of the ambient molecular cloud material . strzer et al . ( @xcite ) point out that the cii layer in a pdr rapidly recombines , but only slowly forms co , leading to a layer with enhanced ci abundance for about @xmath37 years when the illuminating uv radiation is switched off . under repeated exposure to uv , as will naturally occur in a turbulent , clumpy medium where the clumps mutually shadow each other , this will on time average lead to a significantly enhanced ci emission , in particular if the medium stays warm , as will be the case in clouds like the galactic center clouds or the interstellar medium in m82 . which of these proposed mechanisms ( and possibly others ) provide the right explanation for the enhanced ci abundance in m82 and other starburst galaxies has to await further investigation . the measured line ratio allows the determination of an upper limit to the ci column density as long as the emitting gas has temperatures below about 100k . higher temperatures , however , are rather unlikely as there is no known mechanism that would heat the bulk of the gas to such high temperatures and yet keep the dust cool enough to be consistent with the observed dust continuum emission . at temperatures of about 50k the observed line ratio implies an upper limit to the ci column density of about @xmath38 ( with the equivalent width of the line of 100 ) , 20 times larger than the beam averaged column density . the filling factor of individual clumps hence must be about 5% or larger . at these high column densities both lines become close to optically thick . we now argue , that the emitting gas in m82 originates most likely from a large number of small clumps , where the individual clumps reach column densities close to the upper limit discussed above . with _ i ) _ a ci / co abundance ratio of 0.5 , _ ii ) _ a total gas phase abundance of c / h of @xmath39 , _ iii ) _ assuming the gas phase carbon abundance to be dominated by co and ci , and _ iv ) _ an volume density of about @xmath30 estimated from the excitation analysis , a ci column density of @xmath34 corresponds to a geometric depth of the cloud of only 0.5 pc , or 0.03 `` at the distance of m82 of 3.25 mpc ( tamman & sandage @xcite ) . even at the upper range of column densities derived above , i.e. @xmath40 for a giant molecular cloud complex with a 30 wide line , the linear scale reaches only 3.6 pc , or about 0.22 '' . this angular scale is consistent with the beam filling factor estimated above , assuming the emission to arise from on the order of 100 such complexes within the beam . in fact , structure is visible down to at least 2 " on interferometric maps both in co ( shen & lo @xcite ) and in the high density tracing molecule hcn ( brouillet & schilke @xcite ) . higher densities would imply smaller clump sizes and hence a larger number of clumps in the beam . at higher temperatures the upper limit to the column density becomes higher ( by an order of magnitude at 120k ) , and accordingly larger clump sizes are possible . the most likely scenario explaining the observed line emission is thus an ensemble of cloud complexes with linear sizes of a few parsecs and densities of several @xmath30 , their temperatures ranging between about 50k and up to about 100k . in summary , our simultaneous observations of both fine structure lines at 809 and 492 ghz constrain the physical conditions in the -emitting gas to densities of @xmath30 or slightly higher and temperatures of 50k , possibly up to 100k for a fraction of the gas . the analysis confirms the high ci / co abundance ratio of 0.5 in the starburst galaxy m82 . the emission arises in many cloud complexes of a few parsecs extent , which individually reach optical depths close to unity in the lines . this research was supported by the _ verbundforschung astronomie _ through grant 052ku134(6 ) and by the _ deutsche forschungsgemeinschaft _ through grant sfb 301 . special thanks go to members of the smt technical staff for their help during setup of the instrument and interfacing it to the telescope . goldsmith , p.f . , 1982 , quasi - optical techniques at millimeter and submillimeter wavelengths , in : button , k.j . ( ed . ) , infrared and millimeter waves , vol . 6 : systems and components , academic press , new york , 277	we report the first extragalactic detection of the neutral carbon [ ci ] fine structure line at 809 ghz . the line was observed towards m82 simultaneously with the line at 492 ghz , providing a precise measurement of the @xmath0 integrated line ratio of 0.96 ( on a scale ) . this ratio constrains the [ ci ] emitting gas to have a temperature of at least 50 k and a density of at least @xmath1 . already at this minimum temperature and density , the beam averaged ci - column density is large , @xmath2 , confirming the high ci / co abundance ratio of @xmath3 estimated earlier from the 492 ghz line alone . we argue that the [ ci ] emission from m82 most likely arises in clouds of linear size around a few pc with a density of about @xmath4 or slightly higher and temperatures of 50k up to about 100k .
recently , the mn - doped gaas system has received considerable attention in view of its potential use in spintronic technology , as it combines both semiconducting / semimetallic and ferromagnetic properties in one physical system @xcite . the positions of mn dopants play a decisive role in determining the magnetic properties of ( ga , mn)as . there are three of them with a comparable energy @xcite . the substitutional mn atoms , mn@xmath0 , occupying ga sites , act as hole - producing acceptors which contribute to ferromagnetism . the mn atoms in tetrahedral interstitial positions , surrounded by either as or ga atoms , mn@xmath1 and mn@xmath2 , are electron - producing double donors which hinder ferromagnetic states . they partly compensate the mn acceptors in the substitutional positions and reduce the number of holes that mediate ferromagnetism @xcite . post - growth annealing of ( ga , mn)as samples at low temperatures , close to the growth temperature , is known to improve magnetic properties of the ( ga , mn)as system @xcite . the concentration of mn interstitials is reduced via out - diffusion through the hexagonal interstitial sites towards a surface where they oxidize . a mn - rich oxide layer formed on the surface can be removed by etching @xcite . the x - ray absorption spectroscopy is the traditional tool which is used to characterize the local environment of impurities @xcite . not long ago a few works have been published reporting on computations of xanes at the mn k - edge in ternary ( ga , mn)as alloys @xcite and in related diluted magnetic semiconductors @xcite . however , most of these xanes studies concern mainly substitutional mn defects . in the present paper a special attention is paid to interpretation of x - ray absorption spectra measured on mn - doped gaas materials in which different types of impurities coexist . let us note that recently detailed investigations focused on the configuration of substitutional and interstitial mn impurities in ( ga , mn)as have been performed with cross - sectional scanning tunneling microscopy ( xstm ) @xcite and electron paramagnetic resonance ( epr)@xcite . our study is based on the comparison of experimental xanes with spectra calculated by the method of full potential linearized augmented plane waves ( flapw ) @xcite . an isolated defect is represented by a supercell which should be as large as possible to describe correctly the absorption process . therefore , we simulate the spectra from different types of defects by separated calculations , considering always only one substitutional or interstitial atom in a supercell . if we consider an isolated mn impurity , the position of the fermi energy on the energy scale depends on the defect type , i.e. , it is fixed by the acceptor level near the top of the valence band for the substitutional atom and by the donor levels close to the bottom of the conduction band for the interstitial atoms , respectively . the situation is similar in the supercell scheme with a single defect in each supercell . the concentration of defects is finite , the acceptor / donor levels are broadened and merge with the valence / conduction band , but the fermi energy remains fixed near the valence or conduction band , depending on the defect type . this complicates the quantitative analysis of experimental absorption spectra on the basis of the linear combination of theoretical spectra calculated separately for each impurity . in the real crystal with a mixture of defects the position of the fermi energy is determined by the type of defects with the higher concentration , i.e. , in the ferromagnetic ( ga , mn)as by concentration of acceptors . we model this situation by replacing the fermi energy of supercells with donors by the fermi energy of the supercell with an acceptor . this shift of the fermi energy must be respected in the linear combination of spectra fitting of the experimental data . the measurements have been carried out at the european synchrotron radiation facility in grenoble , beamline bm29 . the xanes spectra have been obtained by measuring the mn k@xmath3 fluorescence as a function of the energy of incident photons around the energy of the mn k absorption threshold , @xmath4 ev . the measurements have been performed in the grazing - incidence mode , in which the penetration depth of the primary x - ray beam was sensitively tuned by changing the incidence angle @xmath5 of the primary radiation in the range from 0.1@xmath6 to 0.45@xmath6 , i.e. , around the critical angle @xmath7 of total external reflection . increasing the incidence angle in this range , the penetration depth grows from few nanometers to several microns . the fluorescence method of measurement was employed . the measured fluorescence signal comes from the transition of @xmath8 l - shell core - level electrons to @xmath9 k - shell core holes created by primary x - ray wave . the energy of the corresponding k@xmath3 fluorescence line is characteristic of the mn atom and its intensity is proportional to the absorption coefficient ( the xanes signal ) . spectra were measured at room temperature . the sample grown by molecular beam epitaxy ( mbe ) with @xmath10 of mn content was used in our experiments . after annealing , the content of mn was reduced to @xmath11 . the annealed sample was etched in @xmath12 hcl . manganese k - edge xanes spectra were taken in fluorescence from the 42 nm thick sample of the as - grown , annealed and etched - after - annialing ( ga , mn)as . two sets of experimental mn k - edge xanes spectra , measured on the annealed and annealled - and - etched ( ga , mn)as samples , are presented in fig . [ experiment ] for various incidence angles @xmath5 . ( color online ) experimental k - edge x - ray absorption spectra of mn in ( ga , mn)as measured at different incidence angles @xmath5 of the primary x - ray beam . _ a ) _ spectra taken on the annealed sample with the mn - rich surface oxide layer . _ b ) _ spectra taken on the annealed - and - etched sample without the mn - rich surface oxide layer . all the spectra are normalized with respect to their high - energy ends . the different structures labeled a ( pre - edge ) , b ( pre - edge ) and c ( main absorption peak ) are discussed in the text . ] the spectra can be separated into two parts : the pre - edge and main absorption . the pre - edge region , peaks a and b at around @xmath13 ev , lies below the steeply rising absorption edge , which culminates in the most intense post - edge peak c in the vicinity of @xmath14 ev , as illustrated in fig . [ experiment ] . all experimental spectra , including not presented here as - grown ones , exhibit the double structure ( peaks a and b ) separated by @xmath15 ev in the pre - edge absorption region . in the annealed sample measured at lower @xmath5 , the absorption edge grows almost linearly [ fig . [ experiment]_a ) _ ] . with @xmath5 increased , the spectra in this region acquire a convex shape and the position of the peak c moves to higher energies . the spectra taken at high @xmath5 remind those measured on the annealed - and - etched sample [ fig . [ experiment]_b ) _ ] . this spectrum modification must be due to removing the oxide layer from the sample surface by etching . the theoretical mn k - edge xanes spectra and electronic band structures of ( ga , mn)as were simulated using the full potential linearized augmented plane wave plus local orbitals ( flapw + lo ) method@xcite implemented by blaha et al . in the wien2k package @xcite . a generalized gradient approximation ( gga ) @xcite was employed for the exchange - correlation functional . basis wave functions were expanded in combinations of spherical harmonics inside non - overlapping muffin - tin ( mt ) spheres surrounding the atomic sites and in fourier series in the interstitial region , with a cutoff of r@xmath16k@xmath17 fixed at 7.0 , where r@xmath16 denotes the smallest mt sphere radius and k@xmath17 is the magnitude of the largest @xmath18-vector in the plane wave expansion . the mt radii were assumed to be 1.29 , 1.22 and 1.14 23 a for mn , ga and as , respectively . brillouin zone integrations were performed using a tetragonal @xmath18-point mesh of monkhorst - pack type @xcite . convergence of self - consistency was obtained using @xmath19 @xmath18-point sampling mesh in the reciprocal space of each supercell that corresponds to 10 @xmath18-points in the irreducible wedge of the brillouin zone . the spin - polarized calculations were performed within a 64-atom supercell based on the zinc - blende gaas cubic cell with the experimental lattice constant @xmath20 23 a , doped with 3.125/3.03 atomic percent of mn impurities in the substitutional / interstitial site . we always assumed a single mn atom inside each supercell and used unrelaxed positions of the nearest neighbors around mn . in the substitutional position the mn atom is surrounded by 4 as with the unrelaxed mn - as bond distance 2.445 23 a , the second neighbors are 12 ga on the distance 3.994 23 a. when mn is in one of interstitial positions , its first neighbors are either 4 as or 4 ga , and the second neighbors are either 6 ga or 6 as , respectively , with bond lengths equal to 2.445 23 a and 2.822 23 a. thus , we expect that the difference between the energy structure in two configurations , mn@xmath0 and mn@xmath1 , in which mn is surrounded by 4 as atoms is determined mainly by the second nearest neighbors . the total dos of gaas ( black curve ) together with total doss of ( ga , mn)as with embedded _ a ) _ single substitutional mn atom , _ b ) _ single interstitial mn atom inside the as tetrahedron , _ c ) _ single interstitial mn atom inside the ga tetrahedron , _ d ) _ cluster of four substitutional mn atoms and one interstitial mn atom inside the as tetrahedron , and _ e ) _ cluster from four substitutional mn atoms and one interstitial mn atom inside the ga tetrahedron . the full horizontal line at @xmath21 ev denotes the top of the valence band . the fermi energies of particular supercells are indicated by dashed vertical lines . ] the total density of states ( dos ) calculated for three supercells are shown in the top three panels of fig . [ dos - gaas ] together with the dos of gaas and positions of the fermi levels , which reflect the presence of either acceptor or donor impurities . as mentioned in sec . [ intro ] , the fermi level in ferromagnetic ( ga , mn)as is determined by prevalence of acceptors , and situated in the band composed of the merged valence and broadened acceptor impurity band . consequently , we will consider the states above the mn@xmath0 fermi level as empty also for dos calculated for mn@xmath1 and mn@xmath2 when interpreting xanes spectra . we support this approach by calculation of the fermi level positions in unrealistic 64-atom supercells with incorporated 4 mn@xmath0 and 1 mn@xmath22 . the results are shown in two bottom panels of fig . [ dos - gaas ] . as expected , the resulting fermi energies are shifted closer to the value obtained for the supercell with a single substitutional impurity . in the single - electron approximation , which is used in our model , the absorption spectrum is given by a transition between the initial ground state with fully occupied core - levels and the final state with one core electron removed . the core - hole effect is known to be quite important for reproducing experimental spectra by theoretical calculations @xcite . therefore , we repeated our selfconsistent calculations also for excited states with one core electron removed from the mn @xmath9 level and an additional valence electron placed close to the bottom of the conduction band to fill the lowest unoccupied @xmath23 orbital . the energy of the absorbed radiation is determined as a difference between the calculated total energies of the initial and final states . note , that the influence of mn @xmath9 core - level shifts due to adjacent as or ga atoms is included automatically in the self consistent calculation . for the supercells with interstitial mn atoms this energy must be further corrected by the fermi level shifts , as mentioned above . due to the very small radius of mn @xmath9 k - shell core hole the absorption spectrum is given as a product of the transition probability , calculated from @xmath9 - @xmath23 dipole matrix elements , which is a smooth function of energy , and the local partial dos of @xmath23-states above the fermi energy in the mt sphere surrounding the mn atom , obtained from the final state calculation . normalized ab - initio k - edge xanes spectra of mn in the 64-atom ( ga , mn)as supercell simulated for three diferent mn sites in the substitutional ( mn@xmath0 ) and two tetrahedral interstition positions ( mn@xmath1 and mn@xmath2 ) . solid / dashed curves are spin - polarized spectra calculated with / without the core - hole effect . ] the simulated xanes mn k - edge spectra for three types of isolated mn defects are reported in fig . [ xanes ] . all curves were normalized and convoluted with a lorentz function modeling the experimentally induced broadening . a parameter of a full width at half maximum ( fwhm ) @xmath24 ev was chosen for a fair comparison between experimental and calculated spectra . the use of the broadening with larger @xmath25 would mask the fine details of theoretical curves in the pre - edge region . the transition energy was obtained by a difference in total electronic energies between the ground and core - hole states . the theoretical transition energy deviates from the experimental one ( threshold , @xmath26 ev ) by @xmath27 ev @xmath28 , which is reflected in fig . [ xanes ] by the corresponding shift of the energy scale . similarly to the experimental xanes spectra , the pre - edge and main absorption parts can be distinguished . an intense double structure ( peaks a and b ) appears in the pre - edge region for both types of interstitial defects , mn@xmath1 and mn@xmath2 , in contrast to a weak single peak a for a substitutional mn impurity , mn@xmath0 . arrangement of neighboring atoms around the absorbing mn atom strongly affects the pre - edge structure . the intensity of pre - edge peaks is much larger for interstitials compared to that for the substitutional model spectrum . in accordance with wong @xcite , we attribute this difference to the different distance of the second neighbor shell ligands from the x - ray absorbing mn centers . as already mentioned in sec . [ cm ] , all considered mn impurities are located inside tetrahedrons composed of as / ga atoms with the same bond length of the first neighbor shell ligands . however , the bond distance of the second nearest neighbor ligands of interstitials is by 1.172 23 a less than that of mn@xmath0 . in the wong s terminology , the smaller the `` molecular cage '' , the higher the intensity of the pre - edge absorption . as for the core - hole influence on the shape of the calculated spectra , our results imply that it is significant only in the pre - edge region of the substitutional mn defect and is not important in the main part of absorption curves . with the core - hole effect introduced , the spectral structure of both type of interstitials was only shifted by @xmath29 ev but its shape remained the same . let us now discuss the the pre - edge electronic structure in more detail . mn@xmath30as@xmath31 supercell with a mn atom in the substitutional site calculated without the core - hole effect . _ a ) _ the total doss ( solid curves ) and densities of mn @xmath32-states ( filled regions ) . _ b ) _ the density of @xmath23-states of mn . _ c ) _ the smeared density of @xmath23-states of mn ( the x - ray absorption spectrum divided by the corresponding matrix element ) . the majority / minority spin doss are shown in the upper / lower part of figures . the vertical dash - dot line denotes the fermi level . the electron transitions in the pre - edge part of the absorption spectrum are depicted by arrows . ] calculated with the core - hole effect . ] ( color online ) the same as in fig . [ dos - mn - sub ] for the ga@xmath31mn@xmath30as@xmath31 supercell with the interstitial mn atom inside the as tetrahedron calculated with the core - hole effect . ] ( color online ) the same as in fig . [ dos - mn - sub ] for the ga@xmath31mn@xmath30as@xmath31 supercell with the interstitial mn atom inside the ga tetrahedron with the core - hole effect . ] the band structure of ( ga , mn)as with mn@xmath0 calculated without / with taking into account the core - hole effect is shown in figs . [ dos - mn - sub],[dos - mn - sub - hole ] . when mn atom is in the substitutional position it is surrounded by four as atoms located on the vertices of a tetrahedron . the tetrahedral crystal field of as ions splits @xmath32-states of mn into @xmath33- and @xmath34-levels with @xmath33 below @xmath34 @xcite . exchange interactions further split these states into spin - up ( @xmath35 ) and spin - down ( @xmath36 ) states . due to the tetrahedral arrangement the @xmath34-states of the mn atom hybridize with mn @xmath23-states . the states resulting from hybridization of the deeper @xmath37-states have mainly mn @xmath38 character , while those resulting from higher @xmath39-states have dominantly as @xmath40 character . in contrast , @xmath33-levels of mn have no states available for significant coupling since the gaas host does not have @xmath33-states localized in this energy range . it is seen in fig . [ dos - mn - sub ] that the density of mn @xmath40-states is larger at the energy of @xmath41-states in comparison with that at the energy of @xmath42-states . it is obvious from the discussion above that the pre - edge fine structure of mn k - edge xanes with mn@xmath0 originates from the transfer of @xmath9 electrons to valence @xmath32 states mediated by @xmath9@xmath23 dipole transition and @xmath23@xmath32 hybridization . we observed two weak absorption lines corresponding to @xmath9@xmath37 and @xmath9@xmath39 electron transfers in the absorption spectrum calculated without the core - hole correction [ see fig . [ dos - mn - sub ] _ the first transfer is only partly available as the fermi level touches the @xmath37-band near its border , leaving a little amount of empty @xmath37-states . after introduction of the core - hole effect , two aforementioned lines join into one line of higher intensity [ see fig . [ dos - mn - sub - hole ] _ c ) _ ] , and the transfer @xmath9@xmath39 becames dominant . in this case the mn @xmath37-states are fully filled with electrons and the transfer @xmath9@xmath37 becomes impossible . the band structures of ( ga , mn)as calculated with incorporated interstitials mn@xmath1 and mn@xmath2 , and taking into account the core - hole effect , are shown in figs . [ dos - mn - intas - hole],[dos - mn - intga - hole ] . for both interstitials , we observe two intense pre - edge absorption lines a and b , which are originated from the @xmath9@xmath23 transitions . this is in contrast to the pre - edge structure of mn@xmath0 . both peaks are mainly of dipolar origin . the fermi level separates occupied and unoccupied states . in the case of interstitial impurities it is fixed in the band composed of the merged conduction and broadened donor impurity band . the fermi level falls in the @xmath37-band for mn@xmath1 and @xmath39-band for mn@xmath2 . however , these two bands do not contribute to the pre - edge absorption structure due to low density of mn @xmath32-states near the fermi level . in contrast to the case with the substitutional mn , the @xmath37/@xmath39-states near the fermi level do not participate in @xmath23@xmath32 hybridization . thus , both peaks are mainly of dipolar origin , and their shape replicates the dos of mn @xmath23-states . as already demonstrated in sec . [ exp ] , all experimental spectra ( as - grown , annealed , and annealed - and - etched ) have got the double structure in the pre - edge absorption . in sec . [ fps ] we showed that all examined mn impurities can contribute to the first pre - edge peak a. only mn interstitials are responsible for the intensity growth of the second pre - edge peak b. it implies that at least small proportion of interstitial mn defects is present in the compound . the ratio of experimental pre - edge peak intensities b and a , obtained from measurements on the as - grown , annealed and annealed - and - etched sample as a function of the incidence angle @xmath5 of the primary x - ray radiation . ] the intensity ratios of experimental peaks b and a differs for different incidence angles of the x - ray radiation , @xmath5 . the tendency of b / a variation deep inside the sample is displayed in fig . [ ratio_b_a ] . close to the surface , @xmath43 , the ratio b / a , i.e. , the content of mn atoms in interstitial positions , is smaller in the as - grown and annealed samples , compare to the etched one . it is believed that the as - grown and annealed samples have a thick layer of mn oxide on the surface , and the penetration depth of the x - ray beam inside the sample is small at small @xmath5 , as it is reduced by the width of the oxide layer . in the annealed - and - etched sample the ratio b / a grows slowly with @xmath5 . thus , the mn defect distribution is almost depth independent . the content of mn impurities becomes equal in both annealed and etched samples at large @xmath44 deep inside the sample where the oxid layer is of no importance . comparison of the normalized experimental and theoretical k - edge xanes spectra of mn in ( ga , mn)as . the experimental spectrum ( solid curve ) was measured on the annealed - and - etched sample without the mn - rich surface oxide layer at @xmath45 . the theoretical spectrum ( dashed curve ) is obtained by a linear combination of the model xanes spectra calculated for mn@xmath0 , mn@xmath1 and mn@xmath2 site defects . ] comparison of the normalized experimental and theoretical k - edge xanes spectra of mn in ( ga , mn)as . the experimental spectrum ( solid curve ) was measured on the annealed sample with the mn - rich surface oxide layer at @xmath45 . the theoretical spectrum ( dashed curve ) is obtained by a linear combination of the model xanes spectra calculated for mn@xmath0 , mn@xmath1 , mn@xmath2 site defects and mno oxide . the latter spectrum is shown ( dotted curve ) . ] the simulated data have been processed to explain the experimental xanes spectra . our aim was to guess a rough number of mn impurities that can reproduce the experimental spectra . results are present in figs . [ exp - etch],[exp - anneal ] . before performing a fit we fixed the fermi energy of the supercell with embedded substitutional mn atom , @xmath46 ev . the absorption spectra obtained for mn interstitials were shifted relative to the spectrum calculated for the substitutional mn by @xmath47 , which is @xmath48 ev ( for mn@xmath1 ) and @xmath49 ev ( for mn@xmath2 ) , correspondingly . we found that the content of mn@xmath0 prevails over the content of both interstitial impurities , especially deep inside the sample ( @xmath50 ) . we have got 83@xmath51 of mn@xmath0 and 17@xmath51 of interstitials . the intensity of the peak b can be changed by the variation of the proportion of mn@xmath1 and mn@xmath2 . the more mn@xmath2 , the higher the feature b. analogously to the experimental curve , the theoretical fit has a convex shoulder above the absorption edge , which has been observed only on the annealed - and - etched sample [ fig . [ experiment ] _ b ) _ ] . our fit is not exact which indicates shortages of the component analysis method . the separation of experimental pre - edge peaks a and b is larger by @xmath52 ev in comparison with the distance obtained theoretically . we explain it by the fact that the theoretical width of the energy band gap of the gaas host material is smaller than the experimental width . experimental data obtained for the non - etched annealed sample show an increase in the intensity ratio between the highest peak c and lowest peaks a , b in comparison with the etched sample ( fig . [ exp - anneal ] ) . the shoulder above the absorption edge is observed only for high @xmath5 , the main peak c becomes sharper . such phenomenon can be explained by combined influence of both the content changes of mn defects in ( ga , mn)as and the presence of the oxide layer on the sample surface . to illustrate our assumption , we add an additional component , the mn k - edge xanes spectrum of mno oxide , to the linear combination of three mn model spectra in ( ga , mn)as , and plotted the fit together with the experimental curve , as depicted in fig . [ exp - anneal ] . the main peak c in the fit becomes sharper , the shoulder smooths , and the intensity ratio between the main and pre - edge structures reduces at the expense of the mn oxide component we have performed an extensive lapw numerical study of k - edge xanes spectra of the substitutional and two tetrahedral interstitial mn sites in ( ga , mn)as supercells . we have found out that the effect of the core hole is essential for mn@xmath0 . we studied the whole range of absorption spectra , the highest intensity mn absorption lines in the main part of xanes above the absorption edge and weak lines in he pre - edge part of xanes where the sharp distinction between the spectra with substitutional and interstitial mn defects has been observed . the difference between simulated spectra is determined mainly by the second nearest neighbor ligands . two peaks appear in the pre - edge region for any interstitial defect , whereas a smaller single peak is obtained for a substitutional mn impurity . all pre - edge peaks are small in intensity in comparison with the corresponding highest peak c in the main part . the combination of theoretical spectra for various contents of mn atoms in both substitutional and interstitial positions gives a reasonable qualitative fit of experimental data collected on the annealed and etched sample . summarizing the experimental and theoretical evidence , we have revealed that even after annealing and etching of our sample some mn atoms still reside in interstitial sites , as spectra with two small peaks are always observed experimentally . the assistance of dr . s. pascarelli from the beamline bm28 at esrf grenoble is highly appreciated . this work has been supported by the ministry of education of the czech republic ( center for fundamental research lc510 and the research program msm0021620834 ) , and academy of sciences of the czech republic project kan400100652 . k. w. edmonds , p. bogusawski , k. y. wang , r. p. campion , s. n. novikov , n. r. s. farley , b. l. gallagher , c. t. foxon , m. sawicki , t. dietl , m. b. nardelli , j. bernholc , phys . lett . * 92 * , 037201 ( 2004 ) . p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , j. luitz , wien2k , _ an augmented plane wave plus local orbitals program for calculating crystal properties _ , edited by k. schwarz ( techn . wien , austria , 2007 , isbn 3 - 9501031 - 1 - 2 ) .	this work reports theoretical and experimental study of the x - ray absorption near - edge structure ( xanes ) at the mn k - edge in ( ga , mn)as diluted magnetic semiconductors . the spectra have been calculated from the first - principles using flapw including the core - hole effect , a special attention has been paid to consequences of coexistence of mn impurities in substitutional and tetrahedral interstitial positions . we have performed quantitative component analysis of experimental spectra collected on the ( ga , mn)as samples before / after annealing and etching , with the aim to determine the proportion of mn impurity configurations . comparison of the experimental data with theoretical computations indicates that even after annealing and etching some mn atoms still reside in interstitial sites , although the concentration of interstitial defects has been reduced by annealing .
although globular clusters represent the best available laboratories for constraining stellar evolution models , we now know they are not simple stellar populations . evidence for complex populations are manifested in all phases of stellar evolution . on the main sequence ( ms ) and red giant branch ( rgb ) , high - precision photometry reveals distinct sequences that are most prominent in massive clusters such as @xmath0 cen ( e.g. , anderson 1997 ; bedin et al . 2004 ; ferraro et al . 2004 ) , ngc 2808 ( e.g. , dantona et al . 2005 ; piotto et al 2007 ; milone et al . 2015b ) , m2 ( milone et al . 2015a ) , and ngc 6715 ( e.g. , layden & sarajedini 2000 ; siegel et al . 2007 ; piotto et al . 2012 ) , but the phenomenon is apparently universal ( piotto et al . 2012 , 2015 ) . the formation mechanisms for these multiple populations remain unclear ( see renzini et al . 2015 ) . even before the existence of stellar sub - populations in globular clusters was known , there was considerable evidence for peculiarities in the morphology of the horizontal branch ( hb ) . in particular , there was the well - known `` second parameter '' problem , first mentioned by sandage & wallerstein ( 1960 ; see also sandage & wildey 1967 ; van den bergh 1967 ) ; it refers to the observation that parameters other than metallicity ( such as age and he abundance ) must affect the color distribution of hb stars ( see catelan 2009 for a review ) . in those clusters where the hb distribution is sufficiently broad in color , a series of discontinuities is also evident , although the appearance of these features varies with the bandpasses employed , manifesting themselves as gaps , jumps , overluminous stars , or subluminous stars . three prominent examples of such discontinuities are the `` grundahl jump '' ( g - jump ) within the blue hb ( bhb ) at @xmath111,500 k ( grundahl et al . 1998 , 1999 ) , the `` momany jump '' ( m - jump ) within the extreme hb ( ehb ) at @xmath120,000 k ( momany et al . 2002 , 2004 ) , and the gap between the ehb and `` blue - hook '' stars ( also known as `` late hot flasher '' stars ; dcruz et al . 1996 , 2000 ) , spanning @xmath132,00036,000 k ( sweigart 1997 ; brown et al . 2001 ) . here we have adopted the usual naming convention for hb stars : ehb stars are those at @xmath220,000 k , while bhb stars are those hotter than the rr lyrae instability strip ( i.e. , hotter than @xmath18,000 k ) but cooler than the ehb . both grundahl et al . ( 1999 ) and momany et al . ( 2004 ) noted that their respective jumps appear to be ubiquitous features of globular clusters hosting sufficient numbers of bhb and ehb stars , but the surprising consistency of hb gaps in distinct clusters was recognized somewhat earlier ( ferraro et al . 1998 ) . similarly , the blue - hook phenomenon appears to be common in those clusters hosting an hb population that extends sufficiently far to the blue , but because optical colors become degenerate at the temperatures of ehb stars , uv photometry is needed to confirm their presence ( dcruz et al . 2000 ; brown et al . 2001 ; dalessandro et al . 2008 ; dieball et al . 2009 ; brown et al . 2010 ) . in a review of hot stars in globular clusters , moehler ( 2001 ) explored various explanations for hb discontinuities , including diverging evolutionary paths , mass loss , distinctions in cno or rotation rates , dynamical interactions , atmospheric processes , he mixing in red giants , and statistical fluctuations . with time , it has become increasingly clear that atmospheric processes play a dominant role in these hb features . spectroscopy of stars on either side of the hottest discontinuity demonstrates that , compared to normal ehb stars , blue - hook stars have atmospheres greatly enhanced in he and c ( moehler et al . 2011 ; brown et al . 2012 ) , as expected if they formed through a late he core flash that mixed the h - rich envelope with the inner convective regions ( a process known as flash mixing ; sweigart 1997 ; cassisi et al . the bhb stars hotter than the g - jump exhibit metal abundances enhanced via radiative levitation , he abundances diminished via gravitational settling ( moehler et al . 1999 , 2000 ; behr 2003 ; pace et al . 2006 ) , and lower surface gravities than expected from canonical bhb models ( see moehler 2001 and references therein ) . stellar evolution models that include atomic diffusion have had qualitative success reproducing the observed abundance patterns in bhb stars ( michaud et al . 2007 , 2008 ) . sweigart ( 2002 ) first noted that the onset of radiative levitation on the bhb coincided with the disappearance of surface convection . the interplay between atomic diffusion and surface convection was explored more fully by cassisi & salaris ( 2013 ) ; they noted that by itself , surface convection should not be enough to suppress radiative levitation in stars cooler than the g - jump , implying that other processes , such as turbulence and rotation , must also play a role . in addition to abundance differences , there is a clear bimodality in rotation on the bhb ( recio - blanco et al . 2002 , 2004 ; behr 2003 ) ; stars hotter than the g - jump are generally slow rotators ( @xmath3 sin @xmath4 km s@xmath5 ) , while stars cooler than the g - jump are generally fast rotators ( with @xmath3 sin @xmath6 as high as 40 km s@xmath5 ) . although the source of this bimodality remains unclear , recio - blanco et al . ( 2002 , 2004 ) have speculated that the dearth of fast rotators hotter than the g - jump could be due to a loss of angular momentum through stellar winds , driven by the high atmospheric metallicity at such temperatures . quievy et al . ( 2009 ) have argued that meridional circulation in the fast rotators plays a role in the disruption of atomic diffusion in stars cooler than the g - jump . the consistency of these hb features , and any exceptions to that consistency , are difficult to explore with the heterogeneous observations available in the literature . however , a new catalog of uv and blue photometry , resulting from a _ hubble space telescope ( hst ) _ treasury survey of globular clusters , is well suited to this task ( piotto et al . 2015 ) . in this paper , the seventh in a series associated with the survey , we characterize these hb features in a diverse set of 53 clusters , including clusters with significantly complex populations . we then use these comparisons to explore the implications for the atmospheric phenomena and the evolutionary history of hot stars in globular clusters . our analysis employs photometry obtained with the uvis channel of the wide field camera 3 ( wfc3 ) on board _ hst _ , largely derived from the uv - initiative treasury program 13297 ( pi : piotto ) , but supplemented with data from other guest observer programs ( e.g. , go-12311 , go-12605 ; pi piotto ) . for @xmath0 cen , we use the photometry of bellini et al . ( 2013a , and also in prep . ) , which draws upon archival wfc3 data , including those from calibration programs . the observations employed the f275w ( near - uv ) , f336w ( @xmath7 ) , and f438w ( @xmath8 ) filters ( figure 1 ) the treasury program and data reduction are fully described in piotto et al . ( 2015 ) , and the zero - point calibration and differential reddening corrections are detailed in milone et al . ( 2015a ) . therefore , we will only briefly summarize the observations and data reduction here . images from program 13297 were obtained from aug 2013 to jun 2015 . the data were corrected for charge transfer inefficiency using a pixel - based algorithm developed by anderson & bedin ( 2010 ) for use with the advanced camera for surveys , but later modified for use with the wfc3 . photometry was performed on each individual exposure using a library of spatially - variable empirical point spread functions , combined into a single measurement for each star , and placed in the vegamag system . the primary goal of these imaging programs was to study the multiple populations on the ms and rgb . using observations of ngc 6752 , milone et al . ( 2013 ) demonstrated that the multiplicity of globular cluster populations could be distinguished using the color index @xmath9 ; piotto et al . ( 2015 ) subsequently demonstrated the utility of this index in a large set of clusters . for the cool stars on the ms and rgb , the f275w filter includes an oh molecular band , the f336w filter includes an nh molecular band , and the f438w filter includes cn and ch bands ( milone et al . we will not explore these features here , but they also affect the morphology of the red clump , which falls at similarly cool temperatures . although the program was driven by a desire to characterize the ms and rgb , the same uv data provide insight into hot stars in later evolutionary phases . the hb population is well exposed , with photometric errors of @xmath10.01 mag and nearly 100% completeness . as on the ms and rgb , the @xmath10 color index is useful for characterizing the hb morphology , but for entirely different reasons ( figure 1 ) . at hotter temperatures , the @xmath11 color tracks absorption from fe line blanketing in the near - uv , while the @xmath12 color spans the balmer discontinuity , making the @xmath10 index sensitive to atmospheric metallicity and surface gravity in hb stars . note that these broad - band filters are relatively insensitive to atmospheric he abundance , over the range generally encountered on the hb . specifically , taking the hb spectra at each of the four temperatures shown in figure 1 , if we decrease the atmospheric @xmath13 to 0.01 ( simulating the effects of he gravitational settling ) , or if we increase the atmospheric @xmath13 to 0.40 ( as associated with a self - enriched sub - population ) , the variations within each of our bands are @xmath140.03 mag , and generally @xmath10.01 mag ( see also sbordone et al . 2011 ; girardi et al . 2007 ; dalessandro et al . 2013 ) . in figure 2 , we show the @xmath10 index alongside the color - magnitude diagram ( cmd ) of ngc 2808 , a massive globular cluster long known to host a peculiar hb morphology ( sosin et al . 1997 ; bedin et al . we highlight three discontinuities in the hb distribution : the g - jump ( labeled `` g '' ) , the m - jump ( labeled `` m '' ) , and the gap between the blue - hook stars and the normal ehb ( labeled `` b '' ) . while these breaks can be discerned in the individual colors comprising the @xmath10 index , they are amplified when the colors are combined in this index . furthermore , in these crowded fields , the photometric errors are correlated amongst the various bands ; in the color - color plane ( ccp ) , these errors are suppressed in both axes , while a cmd will suppress them in only one axis . in addition to these three prominent features , there are small gaps ( e.g. , at 0.4 mag and 0.8 mag in @xmath15 color ) , but these do not correspond to features within the hb distributions of the other clusters we will consider here , and could be statistical fluctuations ( see , e.g. , catelan et al . 1998 ) . for reference , the peak in the @xmath10 index corresponds to an effective temperature of @xmath18,600 k , driven primarily by the sensitivity to the balmer discontinuity in the ( @xmath16 ) component of the index . this peak is well - defined in the ccp , falls in the middle of the hb color range , and is populated in nearly all of the clusters in our sample , so we shall use it as a reference point , hereafter called @xmath17 . the zahb models of brown et al . ( 2010 ) that are used in the present paper were computed with a highly - modified version of the original princeton stellar evolution code ( schwarzschild & hrm 1965 ) . this code has been extensively updated over the years ( sweigart & demarque 1972 ; sweigart & gross 1974 , 1978 ; sweigart 1987 , 1997 ) . the equation of state ( eos ) is based on a tabulation of the fermi - dirac integrals and the various thermodynamic functions for both the non - relativistic and relativistic regimes . at low temperatures , the eos solves the saha equation for the ionization of h and he , as well as the first ionization of 10 heavy elements , plus the formation of h@xmath18 . the code also incorporates the opal radiative opacities of rogers & iglesias ( 1992 ) , while at low temperatures , the molecular opacities of bell ( 1995 , private communication ) are used . the nuclear reaction rates are taken from caughlan & fowler ( 1988 ) , except for the @xmath19c(@xmath20 , @xmath21)@xmath22o reaction , where the higher rate suggested by weaver & woosley ( 1993 ) has been adopted . a further discussion of the input physics in our evolution code can be found in the description of the related pgpuc code of valcarce et al . ( 2012 ) . this code has also been highly automated and , as a result , can follow the evolution of a globular - cluster star continuously from the ms up the rgb and then through the he - core flash , hb , and asymptotic - giant branch phases in a single computer run . for example , all of the hb sequences in brown et al . ( 2010 ) were evolved continuously through the he flash , thereby avoiding the need to construct separate zahb models . convective overshooting and semi - convection in our hb models have been treated according to the method of robertson & faulkner ( 1972 ) . this method is applied between iterations of the henyey method and ensures that the radiative and adiabatic gradients agree to better than 10@xmath23 at the convective - core edge and within the semi - convective zone in the final converged models . in addition , the mixing length in our evolution code was calibrated by requiring that a solar model reproduce the solar luminosity and radius as well as the solar @xmath24 ratio at an age of 4.6 gyr . for our analysis , we will use the hb distribution of ngc 2808 as an empirical template for inspecting the hb distributions of the other clusters in the survey . this might seem like an unusual choice , because ngc 2808 is not at all representative of the milky way globular cluster system , being one of the most massive ( @xmath25 mag ; harris 1996 ) , with distinct ms and rgb sequences ( dantona et al . 2005 ; piotto et al . 2007 ; milone et al . 2015b ) and prominent hb gaps ( bedin et al . however , the hb of ngc 2808 is well - populated across the full range of temperature , with stars in the red clump , bhb , ehb , and blue hook ( figure 2 ) . if any of its hb features are the result of universal atmospheric phenomena , as opposed to a product of the multiple populations driving the ms and rgb splitting , then these features should align with those present in the hb distributions of other clusters , as long as the corresponding hb locations are populated . given the prominence of the discontinuities in the ccp of figure 2 , it is useful to compare the hb distribution of each cluster to that of ngc 2808 in this plane . distance is not a factor in a ccp , but extinction is ; we need to align the hb distributions with a fiducial that is independent of the hb discontinuities we are investigating . we use the @xmath17 for this purpose , which is well - populated in nearly all of our clusters . the only exceptions are 9 clusters with hb stars lying entirely in the red clump ; for completeness , we will include them here , and align them to ngc 2808 at the red clump , but their lack of stars hotter than the rr lyrae instability strip makes them irrelevant to the investigation of hb discontinuities . a possible concern when making these empirical comparisons to ngc 2808 is that simple shifts within the ccp are implicitly making the assumption that the extinction vector is independent of spectral energy distribution ( sed ) . because the photometry here covers a broad wavelength range from the near - uv to the @xmath8 band , and because the hb stars in question span a broad temperature range of @xmath15,00040,000 k , this assumption is not entirely correct , but in fact the approximation is sufficient for the purpose of making empirical comparisons between the clusters . this is true even though our survey includes halo clusters with almost no foreground reddening and bulge clusters with significant reddening . the difference in foreground reddening for any cluster of our survey , when compared to that of ngc 2808 , can be as large as 0.5 mag in @xmath26 . to demonstrate the validity of this approximation , we show in figure 3 stellar structure models for the zero - age hb ( zahb ) in ngc 2808 ( brown et al . 2010 ) , which assumed [ fe / h]=@xmath27 ( walker 1999 ) and no he enhancement . ] ( i.e. , @xmath28 = 0 ) , transferred to the ccp using the lte synthetic spectra of castelli & kurucz ( 2003 ) and the wfc3 bandpass throughputs . in one case ( _ solid blue curve _ ) , unreddened synthetic spectra were used to transfer the structure models to the observable plane , and then shifted to align at the @xmath17 . in the other case ( _ dashed red curve _ ) , the structure models were transferred to the observable plane using spectra that were reddened with the fitzpatrick ( 1999 ) extinction curve , assuming @xmath26 = 0.5 mag , and then shifted to align at the @xmath17 . the deviation of the observations below the theoretical zahb curves between the g - jump and m - jump , due to radiative levitation , will be explored extensively below . the two curves agree with each other perfectly in the middle of the hb range ( by definition ) , but separate as one looks to the red and blue extremes of the hb . however , the distinctions between the models are small : at the location of the g - jump , using either the red curve or the blue curve to estimate the effective temperature of the g - jump would only change the estimate by 100 k ( 0.02 mag change in @xmath15 ) , while at the location of the m - jump , the temperature estimates would only differ by 500 k ( 0.04 mag change in @xmath15 ) . thus , even when comparing hb distributions of clusters with significantly distinct foreground reddening , the agreement in the colors of these features implies agreement in their temperatures at the level of @xmath1500 k or less . in cases where the colors of these discontinuities do not agree between clusters , their temperatures can be quantified with a theoretical zahb distribution tailored to match the cluster in question , including an sed - dependent foreground extinction . as a fiducial . we demonstrate the accuracy of this approach by shifting a theoretical zahb distribution to align with the observations at the @xmath17 , using two different extinction assumptions . in one case , no foreground reddening was applied before the alignment ( _ solid blue curve _ ) , and in the other , significant sed - dependent reddening was applied before the alignment ( _ dotted red curve _ ) , but the distinction between the two curves is small . the deviation of the observed bhb below the zahb curves at @xmath29 mag is due to radiative levitation blueward of the g - jump . the stars falling to the blue of the zahb curves , at @xmath30 mag , are blue - hook stars , with effective temperatures hotter than the canonical end of the ehb.,width=326 ] another possible concern when comparing observed hb distributions to that of ngc 2808 is the effect of abundance distinctions between clusters . fortunately , zahb models demonstrate that the intrinsic @xmath15 color is not very sensitive to @xmath13 or [ fe / h ] at the effective temperatures of the @xmath17 ( @xmath18,600 k ) , the g - jump ( @xmath111,500 k ) , or the m - jump ( @xmath120,000 k ) , as demonstrated in figure 4 . in the top panels , we show the effects of [ fe / h ] ; in the bottom panels , we show the effects of @xmath13 . in the left panels , we show the zahb distributions at their relative positions in the ccp , before any alignments are made ; in the right panels , we show the zahb distributions after they have been aligned at the @xmath17 reference point . because ngc 2808 is of intermediate metallicity ( [ fe / h ] = @xmath27 ; walker 1999 ) , using the @xmath17 to align clusters of either high or low metallicity incurs misalignments in @xmath15 color of less than 0.05 mag at the g - jump and m - jump . because the bhb stars near the @xmath17 in ngc 2808 are only moderately enhanced in he ( @xmath31 ; marino et al . 2014 ) , using the @xmath17 to align clusters with little he enhancement or strong he enhancement incurs misalignments in @xmath15 color of less than 0.06 mag at the g - jump and m - jump , which corresponds to a @xmath32 difference of 240 k at the g - jump and 740 k at the m jump . if the systematic errors from [ fe / h ] and @xmath13 went in the same direction , the misalignment could be as large as 0.1 mag . in figure 5 , we compare the hb distribution of ngc 2808 ( _ grey points _ ) to that of each of the other 52 clusters in our sample ( _ blue points _ ) . each of the hb distributions has been aligned to that of ngc 2808 at the @xmath17 , accounting for distinctions in composition and extinction . to aid these comparisons , the 52 clusters have been sorted into five arbitrary categories ( explained below ) with increasingly blue hb morphology . within each category , we also sort the clusters by the mean @xmath33 color of the 10 bluest hb stars in each cluster , after the cluster hb has been aligned to that of ngc 2808 ( otherwise it would depend upon both hb morphology and extinction ) . the choice of sorting metric is arbitrary , but our chosen metric is more useful than other obvious choices , such as the color of the single bluest star ( which is hampered by outliers ) or the mean color of the entire hb ( which is affected by strongly bimodal hb distributions ) . of the 52 clusters compared to ngc 2808 , 17 do not host sufficiently blue hb stars to characterize the three hb discontinuities blueward of the rr lyrae instability strip . of the remaining 36 clusters , all but two ( ngc 6388 and ngc 6441 ) exhibit excellent agreement with the discontinuities observed in ngc 2808 , although the varying hb morphologies of these clusters do not always lend themselves to exploring each feature in detail . furthermore , the red clump in these clusters can vary significantly from that in ngc 2808 , as will be discussed later . our data do not provide sufficient time sampling to identify and characterize rr lyrae stars . however , candidate rr lyrae stars can be flagged by large exposure - to - exposure photometric variations . in figure 5 , we have excluded candidate rr lyrae stars by omitting stars that exhibit photometric variations that significantly exceed those of other stars at the same magnitude ( at the level of 5@xmath34 or greater ) . a representative example of the photometric uncertainty is indicated by an error bar in the upper right - hand corner of each panel . because the rr lyrae instability strip is redder than the discontinuities we explore in this paper , the exclusions do not affect our analysis , but avoid cluttering the ccp with variable stars sampled at random phases . _ category 1 : _ there are 9 clusters dominated by red clump stars ( ngc 6366 , ngc 6838 , ngc 6496 , ngc 6304 , ngc 6652 , ngc 6637 , ngc 6352 , ngc 5927 , and ngc 6624 ) . these clusters provide no information on any of the hb discontinuities . _ category 2 : _ the hb distributions in 8 additional clusters ( ngc 6121 , ngc 6171 , ngc 6584 , ngc 5053 , ngc 4590 , ngc 6397 , ngc 6101 , and ngc 7099 ) extend through the rr lyrae instability strip , but not far enough to populate the vicinities of the grundahl and momany jumps . again , these clusters provide no information on the hb discontinuities . _ category 3 : _ in 6 cases ( ngc 6362 , ngc 5466 , ngc 6981 , ngc 3201 , ngc 1261 , and ngc 362 ) , there are less than four hb stars observed blueward of the g - jump , but those stars do in fact trace the deviations observed in the hb distribution of ngc 2808 . in 4 of these clusters ( ngc 362 , ngc 5466 , ngc 6981 , and ngc 3201 ) there are a few blue - hook stars . however , given the scarcity of bhb and ehb stars , these clusters do not provide interesting constraints on the colors of the hb discontinuities . _ category 4 : _ there are 16 clusters where the region between the grundahl and momany jumps is significantly populated , with few stars hotter than this region . all of the bhb stars trace the hb deviations observed in ngc 2808 , and do so with enough stars to demonstrate consistency in the colors of the grundahl and momany jumps . in 9 cases ( ngc 6717 , ngc 6144 , ngc 5024 , ngc 6535 , ngc 6934 , ngc 6218 , ngc 6341 , ngc 288 , and ngc 6779 ) , there are no stars hotter than the m - jump . in 7 more clusters ( ngc 5272 , ngc 2298 , ngc 6809 , ngc 6723 , ngc 5904 , ngc 6254 , and ngc 4833 ) , those few stars hotter than the m - jump include blue - hook stars . some of these clusters ( e.g. , ncc 5904 , ngc 6779 , ngc 6254 , and ngc 4833 ) , exhibit a sharp decline in the density of hb stars at the m - jump , but this is likely a coincidence , given that other clusters in this category ( e.g. , ngc 6341 and ngc 288 ) exhibit sharp declines unassociated with any particular hb discontinuity . _ category 5 : _ there are 14 clusters ( ngc 6681 , ngc 6656 , ngc 6441 , ngc 5986 , ngc 5286 , ngc 6388 , ngc 6541 , ngc 6093 , ngc 7089 , ngc 6205 , ngc 5139 , ngc 7078 , ngc 6715 , and ngc 2808 itself ) that have significant numbers of stars straddling both the grundahl and momany jumps . most of these also host blue - hook stars , with the exception of ngc 6681 , ngc 6656 , ngc 6441 , and ngc 6093 , which truncate in the gap between the blue - hook stars and the normal ehb population . ngc 6656 ( m22 ) looks somewhat unusual blueward of the m - jump , where the @xmath35 index does not completely overlap with that of ngc 2808 . in most of these 14 clusters , the hb discontinuities have the same @xmath36 colors as those in ngc 2808 ; the exceptions are ngc 6388 and ngc 6441 , which each exhibit a g - jump that is significantly bluer than normal . the red clump distributions of ngc 6388 and ngc 6441 also appear very distinct from those in the other clusters , with ngc 6388 having a bifurcated structure ( see also bellini et al . + + our sample includes 21 clusters with blue - hook stars , extending to 23 the tally of globular clusters known to host blue - hook stars . in table 1 , we list these clusters , along with the total number of hb stars in each sample ( @xmath37 ) , and the total number of blue - hook stars ( @xmath38 ) that are blue enough to be unambiguously classified . specifically , we classify them as blue - hook stars if they fall within @xmath39 mag and @xmath40 mag in the ccp after alignment with ngc 2808 . note that there are clusters in our sample where a few additional stars fall immediately to the red of this selection region , which might also be blue - hook stars , but their classification would not be secure ( ngc 5139 , ngc 5286 , ngc 5986 , ngc 6205 , ngc 6254 , ngc 6541 , ngc 6715 , ngc 6723 , and ngc 7089 ) . rrr\|rrr\|rrr ngc & @xmath37 & @xmath38 & ngc & @xmath37 & @xmath38 & ngc & @xmath37 & @xmath38 + 362 & 303 & 2 & 5286 & 386 & 2 & 6541 & 228 & 3 + 2298 & 56 & 2 & 5466 & 30 & 1 & 6715 & 916 & 120 + 2808 & 757 & 52 & 5904 & 132 & 2 & 6723 & 150 & 1 + 3201 & 31 & 1 & 5986 & 328 & 3 & 6809 & 35 & 2 + 4833 & 138 & 1 & 6205 & 244 & 4 & 6981 & 64 & 1 + 5139 & 825 & 156 & 6254 & 86 & 1 & 7078 & 390 & 7 + 5272 & 179 & 1 & 6388 & 953 & 2 & 7089 & 515 & 3 + to date , uv photometry had been used to confirm blue - hook stars in six globular clusters : ngc 5139 ( @xmath0 cen ) , ngc 2808 , ngc 6715 ( m54 ) , ngc 2419 , ngc 6388 , and ngc 6273 ( dcruz et al . 2000 ; brown et al . 2001 ; dalessandro et al . 2008 ; dieball et al . 2009 ; brown et al . 2010 ) . although our sample does not include ngc 2419 and ngc 6273 , it does include the others , and the ccp of figure 5 confirms the presence of blue - hook stars in each cluster . dieball et al . ( 2009 , 2010 ) provided four additional clusters that may host a small number of blue - hook stars each : ngc 6093 ( m80 ) , ngc 6681 ( m70 ) , ngc 7078 ( m15 ) , and ngc 6441 . of these clusters , three ( ngc 6093 , ngc 6441 , and ngc 6681 ) do not however appear to host blue - hook stars in our sample , but ngc 7078 clearly does . we also find blue - hook stars in 16 additional clusters : ngc 362 , ngc 2298 , ngc 3201 , ngc 4833 , ngc 5272 ( m3 ) , ngc 5286 , ngc 5466 , ngc 5904 ( m5 ) , ngc 5986 , ngc 6205 ( m13 ) , ngc 6254 ( m10 ) , ngc 6541 , ngc 6723 , ngc 6809 ( m55 ) , ngc 6981 ( m72 ) , and ngc 7089 ( m2 ) . in 7 of these clusters ( ngc 3201 , ngc 4833 , ngc 5272 , ngc 5466 , ngc 6254 , ngc 6723 , and ngc 6981 ) , there is only a single star falling unambiguously in the blue - hook region . the classification seems secure , given the placement in both the cmd and ccp , and the lack of other stars in the vicinity that would otherwise suggest significant contamination . in the 23 clusters hosting blue - hook stars , these stars comprise up to 20% of the hb population , although the percentage is highest in three of the most massive clusters ( ngc 2808 , ngc 5139 , and ngc 6715 ) . our sample of 53 clusters spans @xmath41 mag in total luminosity ( harris 1996 ) , with a median of @xmath42 mag ; 17 of the 21 clusters in our sample hosting blue - hook stars are in the brightest half of the sample , reinforcing the idea that blue - hook stars tend to form in the most massive clusters ( see brown et al . 2010 for a full discussion ) . for some of these clusters , the statistics are poor , with only one or two blue - hook stars , but we stress that the presence of blue - hook stars is not simply a matter of sampling enough stellar mass to find a relatively rare evolutionary phase . for example , the mean metallicities of ngc 5139 and ngc 5986 match within 0.1 dex , and the counts of their hb stars in our catalogs match within 3% , but ngc 5139 , being the more massive cluster , has @xmath4320 times as many blue - hook stars as ngc 5986 ; this is another manifestation of parameters beyond metallicity affecting the hb morphology ( i.e. , the second - parameter problem ) . in figure 5 , only two globular clusters exhibit significant discrepancies with the discontinuities of ngc 2808 : ngc 6388 and ngc 6441 . specifically , in both of these clusters , the g - jump is @xmath10.4 mag bluer than that observed elsewhere . in ngc 6388 , the region between the g - jump and m - jump is well populated , such that the shift is obvious . in ngc 6441 , the shift is less statistically significant , with only 6 stars falling blueward of the point where the g - jump occurs in ngc 2808 , and only 2 of these stars at significantly bluer colors . that said , all 6 of these stars are well aligned with the @xmath44 trend extending from the cooler hb stars in ngc 6441 , and over the @xmath36 color range of these 6 stars , there are no stars exhibiting a deviation from this trend ( i.e. , these 6 stars do not trace the locus of ngc 2808 ) . in both ngc 6388 and ngc 6441 , the shift in this feature implies that the onset for radiative levitation occurs at hotter effective temperature than the onset in other clusters , even after one accounts for the composition effects demonstrated in figure 4 . to quantify this shift , in figure 6 we compare the photometry of four clusters ( ngc 2808 , ngc 6715 , ngc 6388 , and ngc 6441 ) to theoretical zahb distributions from brown et al . these zahb models assume that the hb stars evolved from ms stars with the standard chemical composition for each cluster ( i.e. , no he enhancement ) . the stellar structure models were then transferred to the observed ccp using the lte synthetic spectra of castelli & kurucz ( 2003 ) , applying the fitzpatrick ( 1999 ) reddening law to provide appropriate sed - dependent extinction at each point along the zahb ( thus avoiding the small approximation errors demonstrated in figure 3 ) , and subsequently folding the spectra through the wfc3 bandpasses . as in figure 5 , we aligned the theoretical distributions at the @xmath17 , using a least squares fit of the model to the data . the observed hb distribution ( _ black points _ ) between the grundahl and momany jumps clearly deviates from the theoretical distribution ( _ green curves _ ) in figure 6 ; the model is shown as a dashed line over the region where the data deviate from the model . over the range of the observed deviation , we also show the same stellar structure models , but transferred to the observable plane using synthetic spectra with an enhanced metallicity of [ fe / h ] = 0.5 ( the maximum metallicity available in the grid of synthetic spectra ) , simulating the effects of radiative levitation in the atmospheres ( _ blue curves _ ) . for ngc 6388 and ngc 6441 , the deviation observed between the g - jump and m - jump is more significant than that in the model . this may be due to the fact that the stars in these clusters were born at much higher metallicity than ngc 2808 , such that a simulation of radiative levitation would require a metallicity exceeding [ fe / h ] = 0.5 . of course , our use of [ fe / h ] = 0.5 synthetic spectra is only a crude approximation of the actual interplay between gravitational settling and radiative levitation in the stellar atmospheres , which produces large element - to - element variations . the effective temperatures for the g - jump and m - jump in each cluster are determined by observing where the observed hb distribution deviates from the zahb distribution transferred with synthetic spectra at standard cluster composition ( _ green curves _ ) . although the m - jump is well - defined in ngc 2808 , it is spread over a significant color range in ngc 6715 , ngc 6388 , and ngc 6441 . nonetheless , a transition temperature of 20,000 k is consistent with the observed m - jump in each cluster . for ngc 2808 and ngc 6715 , the temperature of the g - jump is close to the @xmath111,500 k temperature reported at the time of its original discovery ( grundahl et al . 1998 , 1999 ) . in contrast , the g - jump occurs at @xmath114,000 k in ngc 6388 and @xmath113,500 k in ngc 6441 , although given the small number of stars in the vicinity of the ngc 6441 g - jump , this value is uncertain at the level of a few hundred degrees . although the models in figure 6 employ metallicities appropriate for each cluster , these clusters exhibit variations in he enhancement along the hb , and the assumed @xmath13 has a small but non - negligible effect on the alignment of the models to the data ( see figure 4 ) . for this reason , in figure 7 we also characterize the effective temperatures of the discontinuities using zahb models representing sub - populations with strong he enhancement ( @xmath28 = 0.17 ) . the temperature determinations shown in figures 6 and 7 bracket the possibilities in each cluster . even if one tries to minimize the temperature distinctions in the g - jump for each cluster , by assuming the coolest estimates for the metal - rich clusters ( ngc 6388 and ngc 6441 ) and the hottest estimates for the clusters at lower metallicity ( ngc 2808 and ngc 6715 ) , the metal - rich clusters have g - jump temperatures at least 1,100 k hotter than those in the other clusters . figures 6 and 7 simulate the effects of radiative levitation in bhb stars by assuming a super - solar metallicity between the g - jump and m - jump . the evidence for radiative levitation blueward of the g - jump is well supported by spectroscopy of bhb stars ( e.g. , moehler et al . 1999 , 2000 ; behr 2003 ; pace et al . 2006 ) . in the ccp , the observed hb locus returns to alignment with the standard zahb distribution ( i.e. , with no atmospheric enhancement of metallicity ) blueward of the m - jump , but the complete cessation of radiative levitation can not be the origin of the m - jump ; instead , some other effect is likely counterbalancing the deviation associated with radiative levitation . there are few metallicity measurements in globular clusters for ehb stars ( i.e. , at @xmath45 k ) , although brown et al . ( 2012 ) found that some ( but not all ) of these stars in ngc 2808 exhibited super - solar fe abundances . the sdb stars of the galactic field population are the analogs of the ehb stars in globular clusters , and geier et al . ( 2010 ) found that the fe enhancement in sdb stars hotter than 20,000 k is similar to that in the bhb population of globular clusters . furthermore , blueward of the g - jump , the enhancement of atmospheric metals is accompanied by a corresponding depletion in atmospheric he . in @xmath0 cen , this he depletion continues at temperatures well past 20,000 k ; in fact , the depletion continues until @xmath46 32,000 k , where the blue - hook stars exhibit atmospheres greatly enhanced in he ( moehler et al . a similar result was found in ngc 6752 , with surface he significantly depleted over the entire range of 12,000 k 32,000 k ( moni bidin et al . 2007 ) . besides this evidence , we can demonstrate with our own data that the m - jump can not be induced by the restoration of normal atmospheric abundances at temperatures above 20,000 k. in figure 8 , we show the same data and models that appeared in figures 6 and 7 , but now the ordinate uses @xmath47 , similar to the @xmath7 band employed by grundahl et al . ( 1999 ) and momany et al . ( 2002 ) in their cmds characterizing the g - jump and m - jump . as noted previously , the use of [ fe / h ] = 0.5 spectra is a crude approximate for the abundance variations incurred through atmospheric diffusion , and the hb exhibits a range of @xmath13 values at any particular color , due to dispersions in rgb mass loss . it is impossible to disentangle these complexities using broad - band photometry . however , the purpose of the comparison here is to demonstrate the qualitative effects of radiative levitation in the atmosphere ( and the effects of he abundance in the stellar structure models , which will be discussed below ) . the onset of radiative levitation blueward of @xmath48 mag causes an upward jump in the @xmath47 photometry , such that the sense of the shift is the same in the model and the data . however , the cessation of radiative levitation blueward of @xmath49 mag causes a downward shift in the model , in contrast to the data . thus , the onset of radiative levitation explains the g - jump in both the ccp ( figures 6 and 7 ) and the cmd ( figure 8) , but the cessation of radiative levitation does not simultaneously reproduce the behavior of the m - jump in both diagrams . although a discussion of the red clump morphology in these clusters is beyond the scope of the current paper , we note some clusters exhibit red clump distributions in the ccp ( figure 5 ) that are distinct from that of ngc 2808 . the distinction can be modest ( e.g. , ngc 5286 and ngc 5904 ) or severe ( e.g. , ngc 6388 and ngc 6441 ) , and is likely driven by metallicity effects in the reddest hb stars . however , even if we calculate zahb distributions at the appropriate metallicity for ngc 6388 and ngc 6441 , there is a significant mismatch between the models and data in figures 6 and 7 . these stars are as cool as those on the ms , where the sensitivity of the @xmath50 index to cno abundances makes it a useful diagnostic in the exploration of multiple populations ( see milone et al . the distortions of the red clump here may be related to the distinct ms morphologies of these two metal - rich clusters . bellini et al . ( 2013b ) demonstrated that the ms of ngc 6441 is clearly split into two branches , while that of ngc 6388 is broadened but not split . they hypothesized that the second generation of stars in each cluster has a similarly enhanced he abundance but distinct cno abundances . those cno variations may explain why the two clusters exhibit distinct red clump morphologies in figures 6 and 7 . the predicted morphology of hb evolutionary tracks with cno - enhanced mixtures ( pietrinferni et al . 2009 ) seems to be consistent with this possibility ( see also the optical analysis of ngc 1851 by gratton et al . 2012 ) . the g - jump is associated with a sharp increase in atmospheric metallicity , decrease in atmospheric he , and decrease in stellar rotation . we can now add that the g - jump is almost universally consistent in effective temperature , with the notable exceptions of ngc 6388 and ngc 6441 . the most likely reason for the increased temperature of the g - jump in ngc 6388 and ngc 6441 is that the bhb stars in these two metal - rich clusters are significantly enhanced in he , compared to bhb stars in relatively metal - poor clusters ; at higher metallicities , a larger he abundance is needed to populate the bhb . as mentioned previously , all globular clusters appear to exhibit sub - populations with distinct chemical compositions , but the phenomenon is strongest in massive globular clusters , where there exist sub - populations enhanced in he up to @xmath51 ( @xmath52 ; see piotto et al . 2015 and references therein ) . at a fixed age , the ms turnoff mass decreases as he increases , and thus for a given amount of rgb mass loss , ms stars at higher @xmath13 tend to produce hb stars of lower mass and higher effective temperature . he enhancement also affects the hb luminosity . on the red clump and bhb , he - enhanced stars are brighter than normal , due to the larger energy output of their hydrogen shells that results from their lower envelope opacity and higher envelope mean molecular weight ( see , e.g. , sweigart 1987 ) . in contrast , on the ehb , he - enhanced stars are fainter than normal , since they have smaller he core masses ( see , e.g. , sweigart & gross 1978 ) and insufficient envelope masses to support an active hydrogen shell ( valcarce et al . ngc 2808 is an instructive example for these effects . from the first parameter of hb morphology ( metallicity ) , one might expect an hb distribution that does not extend to the end of the ehb , given its intermediate metallicity ( [ fe / h ] = @xmath27 ; walker 1999 ) = @xmath53 on the uves scale , but the distinction makes no difference in our analysis here . ] . however , we know that massive clusters tend to host significant ehb populations associated with he - enriched populations ( see milone et al . 2014 ) . this is true in ngc 2808 ( @xmath54 mag ; harris 1996 ) , which hosts sub - populations with varying amounts of he enhancement ( up to @xmath51 ; dantona et al . 2005 ; piotto et al 2007 ) ; its hotter hb stars are generally drawn from sub - populations with higher @xmath13 , with increasingly large deviations from the luminosity of the canonical hb ( e.g. , dantona & caloi 2004 ; brown et al . 2010 ; dalessandro et al . 2011 ) . in the atmospheres of its bhb stars , marino et al . ( 2014 ) also found that he is enhanced , until the point of the grundahl jump , where he is depleted through gravitational settling . the varying progenitor populations for hb stars as a function of color is reinforced by the work of gratton et al . ( 2011 ) , who found that the bhb stars in ngc 2808 are o - poor and na - rich ( corresponding to the blue sub - population on the ms ) , while its red hb stars are o - rich and na - poor ( corresponding to the red sub - population on the ms ) . another example of these effects can be seen in the comparison of m3 and m13 , two intermediate - metallicity clusters with distinct hb morphology that can be traced to he enhancement ( @xmath550.04 ) in m13 ( dalessandro et al . 2013 ) . along similar lines , the metallicities of ngc 6388 ( [ fe / h ] = @xmath56 ; piotto et al . 2002 ) and ngc 6441 ( [ fe / h ] = @xmath57 ; harris 1996 ) are so high = @xmath58 and @xmath59 , respectively , on the uves scale , but as with ngc 2808 , the distinction makes no difference to our analysis here . ] that they would normally produce an hb falling entirely in the red clump . however , the hb morphology of each cluster extends far to the blue ( rich et al . 1997 ) , including significant populations of unusually - bright rr lyrae stars ( layden et al . 1999 ; pritzl et al . 2001 , 2002 , 2003 ; corwin et al . 2006 ) that belong to neither oosterhoff class ( pritzl et al . the hb of each cluster is over - luminous blueward of the red clump , implying that the bhb stars originate in a ms population enhanced to @xmath51 ( or equivalently @xmath60 0.140.17 ; busso et al . 2007 ; caloi & dantona 2007 ; dantona & caloi 2008 ; brown et al . he enhancement is required to move a metal - rich hb star blueward from the red clump to the bhb while increasing its luminosity . to demonstrate the effects of @xmath13 enhancement on hb luminosity , we show in figure 8 the same hb distributions of figures 6 and 7 , but using cmds with @xmath47 ( @xmath7 ) on the ordinate . to ease comparisons between the clusters , the observed hb distributions are again aligned to that of ngc 2808 ( with the same color alignment employed in previous figures ) . here , the theoretical zahb distributions ( brown et al . 2010 ) have been calculated for stars at @xmath61 ( _ dark green _ ; the same models in figure 6 ) and @xmath62 ( _ light green _ ; the same models in figure 7 ) . as in figures 6 and 7 , the radiative levitation of metals is simulated through the use of super - solar spectra ( _ blue _ ) . the abscissa alignment of the models to the data is the same as that used in figures 6 and 7 , while the ordinate alignment of the models to the data places the zahb model with @xmath63 at the base of the observed red clump ( on the assumption that the faintest red clump stars arise from a population unenhanced in he ) . in ngc 6715 , the g - jump is clearly visible , but to the red of the g - jump , the observed hb stars are closer to the model for stars at @xmath61 , implying that hb stars in the vicinity of the g - jump were born with little he enhancement . in ngc 2808 , the g - jump is again clearly visible , but to the red of the g - jump , there is a more significant luminosity difference between the observations and the model for stars at @xmath61 , implying that the hb stars in the vicinity of the g - jump were born with some he enhancement . for reference , dantona & caloi ( 2004 ) used photometry to estimate that these stars have moderate enhancement , with @xmath600.040.06 ( see also dalessandro et al . 2011 ) , while marino et al . ( 2014 ) used spectroscopy to measure a he enhancement of @xmath64 ( internal plus systematic uncertainty ) . for ngc 6388 and ngc 6441 , the hb stars near the g - jump are much more luminous than the model for @xmath63 , and nearly as bright as the model for @xmath62 . this is in agreement with previous photometric results ( e.g. , busso et al . 2007 ; caloi & dantona 2007 ; dantona & caloi 2008 ; brown et al . 2010 ; bellini et al . furthermore , the g - jump appears more complicated for these two clusters . for ngc 6441 , the paucity of stars in the vicinity of the g - jump makes the jump difficult to discern ( at least compared to the clear deviation in figures 6 and 7 ) . for ngc 6388 , however , there appears to be a pair of jumps : one at the usual temperature for the g - jump ( i.e. , near 11,500 k ) , and a hotter one ( i.e. , near 14,000 k , corresponding to the obvious deviation in figures 6 and 7 ) ; these appear as excursions from the dominant stellar locus . if the bhb stars of ngc 6388 and ngc 6441 were born with a large he enhancement , can this explain the increased temperature for the g - jump in these clusters ? to investigate this point , we show in figure 9 the behavior of the convective zones due to h , , and ionization near the surface of hb stars , as a function of effective temperature , using evolutionary models calculated at metallicities appropriate for ngc 6388 and ngc 2808 . the figure shows the convective zones for hb stars with @xmath65 ( _ red shading _ ) and @xmath62 ( _ blue shading _ ) . sweigart ( 2002 ) and cassisi & salaris ( 2013 ) used similar figures ( with no he enhancement ) to demonstrate how the surface convection from the ionization zone disappears at temperatures hotter than @xmath112,000 k , enabling the onset of radiative levitation in bhb stars and the appearance of the g - jump . here , we show the convection zones for two different @xmath13 assumptions , demonstrating that the convection is sensitive to initial @xmath13 , while the and h convection zones are not . comparing the convective zones in the top ( intermediate metallicity ) and bottom ( high metallicity ) panels , it is clear that @xmath13 ( and not [ fe / h ] ) is the critical parameter governing the behavior of the convection zone , and thus the g - jump , although we note that there are other changes in the convection zones that depend upon [ fe / h ] . specifically , increasing [ fe / h ] shifts the termination of the h convection zone hotter by 5001000 k ( depending upon @xmath13 ) and extends the tail of the convection zone to both hotter temperatures and shallower depths . ( _ red shading _ ) to @xmath62 ( _ blue shading _ ) , the transition in the convective zone shifts to higher effective temperature by @xmath12,000 k , and the convective zone moves closer to the surface . if the g - jump is normally associated with the convective zone , this temperature shift at high @xmath13 may explain the bluer g - jump in ngc 6388 and ngc 6441 . if the g - jump is normally associated with the h convective zone , the surface encroachment of the convective zone at high @xmath13 may also explain the bluer g - jump in ngc 6388 and ngc 6441 . note that evolutionary effects , turbulence , and mass loss complicate this interpretation ; these models only serve to provide possible explanations for the observed g - jump behavior in a qualitative sense . similarly , if the m - jump is associated with the convective zone , the observed consistency of the m - jump temperature ( even in ngc 6388 and ngc 6441 ) might be due to the insensitivity of the convective zone to @xmath13 . , width=326 ] with this behavior in mind , the convection zones in the vicinity of the g - jump ( from h and from ) offer two possible explanations for the g - jump . qualitatively , a convective zone acts as a fully mixed reservoir of matter having the original chemical composition , which minimizes the effects of diffusion ( gravitational settling and radiative levitation ) . the g - jump could be due to the ionization , as hypothesized by sweigart ( 2002 ) and cassisi & salaris ( 2013 ) . in this view , the bhb stars in most clusters are born with little to modest he enhancement , such that there is little variation in the effective temperature of the g - jump . in ngc 6388 and ngc 6441 , the bhb stars have a significant enhancement near @xmath66 ; along the hb , this shifts the cessation of ionization to hotter effective temperatures by @xmath12,000 k the same shift observed . in this scenario , it is unclear if the g - jump in ngc 2808 ( with stars near @xmath67 ) would be so similar to the g - jump in less massive clusters that exhibit little he enhancement in their sub - populations ( see figure 5 ) . alternatively , the g - jump in most clusters could be due to the h convection zone . in this view , the g - jump remains at constant effective temperature in most clusters because the h convection is insensitive to @xmath13 . however , for bhb stars with he abundances near @xmath66 , the ionization shifts to hotter effective temperature and moves closer to the stellar surface . this would make the g - jump appear at hotter effective temperature ( as in figure 5 ) if it arises from convection , or even multiple temperatures ( as might be implied by figure 8) if it arises from both h and convection . with both of these scenarios , the temperatures observed for the g - jump do not exactly coincide with the transitions in the modeled convection zones , even if the behavior is qualitatively consistent . it is likely the case that other parameters , such as turbulence , play a role in the exact location of the jumps ; the region that is mixed at the surface may not coincide with the formal convective boundary . along these lines , we note that michaud et al . ( 2011 ) reproduce the g - jump at 11,500 k by invoking a fully mixed region near the surface of the star , with a mass 10@xmath68 @xmath69 , possibly driven by turbulence or mass loss . another complication concerns the direction of the hb evolution beyond the zahb . as known since the work of sweigart & gross ( 1976 ) , bhb stars with normal he abundances can slowly evolve redward toward cooler effective temperatures , especially at low to intermediate metallicities . however , when their he enhancement is @xmath62 , the hb stars in the vicinity of the g - jump can rapidly evolve blueward from temperatures cooler than 11,500 k to temperatures near 15,000 k. depending upon the relative timescales of the various factors at work ( surface convection , turbulence , mass loss , radiative levitation , gravitational settling ) , the distinct evolutionary paths for high-@xmath13 stars may also push the g - jump toward higher effective temperatures in clusters like ngc 6388 and ngc 6441 . an exploration of these effects is currently underway ( tailo et al . , in prep . ) . as noted previously , the m - jump can not be due to a simple disruption of radiative levitation at temperatures hotter than 20,000 k. looking at figure 9 , it is worth noting that the convection zone begins to encroach upon the surface near this temperature , and the behavior is independent of he abundance . it may be a coincidence that the m - jump is also independent of he abundance , even in the two clusters exhibiting a hotter g - jump ( ngc 6388 and ngc 6441 ) , but we speculate that the m - jump may be associated with the convection zone . using uv and blue photometry for 53 galactic globular clusters , we have shown that the discontinuities in their hb distributions are remarkably consistent . globular clusters are now known to host complex populations with variations in chemical composition , but these hb discontinuities reflect universal transitions in atmospheric phenomena , and not abundance distinctions in their ms progenitors . that said , the effective temperature for one of these discontinuities , the g - jump , is @xmath11,0002,000 k hotter in ngc 6388 and ngc 6441 . this shift is likely due to the fact that these two clusters host bhb stars greatly enhanced in he ( @xmath52 ) , which affects the behavior of the convective zone and its role in disrupting radiative levitation . although the complexity of globular cluster populations was originally recognized in the most massive globular clusters , its ubiquity became more apparent with appropriate photometry . the history of blue - hook stars is following a similar path . this is not because massive clusters provide more chances to find a star following a relatively rare evolutionary avenue ; instead , these clusters host sub - populations significantly enhanced in he , which leads to a hotter hb morphology . while blue - hook stars comprise a tiny fraction of the population in any globular cluster , we have shown that these products of extreme mass loss can be found in most of the bright globular clusters . the census of clusters hosting blue - hook stars has increased to 23 , with nearly all of them residing in the brighter half of the galaxy s globular cluster system . support for program go-13297 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc . , under nasa contract nas 5 - 26555 . s.c . and g.p . recognize partial support by the iac ( grant p301031 ) and the ministry of competitiveness and innovation of spain ( grant aya2010 - 16717 ) ; s.c . , g.p . , and f.d . recognize partial support by prin - inaf 2014 ( pi : s. cassisi ) . s.o . and g.p . also acknowledge partial support by the universit degli studi di padova progetto di ateneo cpda141214 `` towards understanding complex star formation in galactic globular clusters . '' a.p.m . acknowledges support by the australian research council through discovery project grant dp120100475 . f.r.f . and e.d . acknowledge the support from the cosmic - lab project ( web site : http://www.cosmic-lab.eu ) funded by the european research council , under contract erc-2010-adg- 267675 . a.p . acknowledges support from prin - inaf2012 ( pi : l. bedin ) . s.o . gives thanks for the support of the university of padova .	the uv - initiative _ hubble space telescope _ treasury survey of galactic globular clusters provides a new window into the phenomena that shape the morphological features of the horizontal branch ( hb ) . using this large and homogeneous catalog of uv and blue photometry , we demonstrate that the hb exhibits discontinuities that are remarkably consistent in color ( effective temperature ) . this consistency is apparent even among some of the most massive clusters hosting multiple distinct sub - populations ( such as ngc 2808 , @xmath0 cen , and ngc 6715 ) , demonstrating that these phenomena are primarily driven by atmospheric physics that is independent of the underlying population properties . however , inconsistencies arise in the metal - rich clusters ngc 6388 and ngc 6441 , where the discontinuity within the blue hb ( bhb ) distribution shifts @xmath11,000 k to 2,000 k hotter . we demonstrate that this shift is likely due to a large helium enhancement in the bhb stars of these clusters , which in turn affects the surface convection and evolution of such stars . our survey also increases the number of galactic globular clusters known to host blue - hook stars ( also known as late hot flashers ) from 6 to 23 clusters . these clusters are biased toward the bright end of the globular cluster luminosity function , confirming that blue - hook stars tend to form in the most massive clusters with significant self - enrichment .
in the main text , we discussed how the `` parity anomaly '' gives rise to a change in landau level counting in magnetic field and discussed how this fact is related to signatures in the magnetotransport . for this purpose , we compare the continuity equation , derived from the semiclassical equations of motion , with the continuity equation , derived from maxwell equations , and show that the qah level has a non - zero chern number but an effective zero charge carrier density . therefore , in the process of adiabatically switching on magnetic field , the qah state can be distinguished from a trivial insulator when the chemical potential is in the mass gap . the structure of this appendix is the following : in the first part , we give an explicit derivation of the `` parity anomaly '' based on a semi - classical formulation , while in the second part we show how an additional topological term enters the maxwell equations . breaking of parity symmetry due to the dirac mass term is discussed in appendix b. let us start from the spin - up block of the bhz hamiltonian @xcite , @xmath143 with @xmath144 , where we have neglected the @xmath145-dependence of the mass term . this approximation is valid in the low energy limit @xmath146 which means at least up to an energy of @xmath147 . following , we show that the parity - breaking dirac mass gives rise to an anomalous transverse current response to an applied electric field , as well as to a non - zero carrier density in the ground state . we demand the chemical potential to be located within the mass gap , so that the bulk is insulating and is characterized by a non - zero chern number @xmath148 where @xmath149 is the berry curvature and @xmath150 are eigenstates of the dirac hamiltonian . in a semi - classical approach the berry curvature acts as magnetic field in @xmath151-space and induces an anomalous correction to the velocity @xcite . in order to conserve the phase space volume in magnetic field , it must also enter as a correction in the density of states , whereby the carrier density is given by @xcite @xmath152 where @xmath153 is a background carrier density and we demand that @xmath154 ( bulk insulator ) . based on this result , it is straightforward to derive the transverse hall conductance from streda s formula @xcite , @xmath155 the result of eq . ( [ eqa : densityresponse ] ) and ( [ eqa : hallresponse ] ) can be combined in a covariant form @xmath156 where @xmath157 ( @xmath158 ) is the charge carrier density and @xmath159 is the electromagnetic tensor . in comparison to the qh effect , where the current originates from the magnetic field and has a @xmath160-dependence , the anomalous current is hereby induced by the dirac mass term . since the mass term breaks parity , this effect in solid state physics is known as `` parity anomaly '' . if the mass term changes sign across a topological domain wall as shown in fig . [ fig : qhvsqah ] , charge conservation is violated at the interface , corresponding to a ( 1 + 1)d system , known as chiral anomaly , @xmath161\epsilon_{\mu\nu}f^{\mu\nu}\\ & = & \eta \frac{e^2}{2h}\epsilon_{\mu\nu}f^{\mu\nu}\label{eqa : chiralanomaly},\end{aligned}\ ] ] where @xmath162 /2 $ ] . while the fermion - doubling theorem ensures in even spacetime dimensions that a second domain wall ( fermion ) must exist which cancels this anomaly in total , there is no such theorem in odd spacetime dimensions @xcite . since the dirac mass term breaks parity , there is an intrinsic chern - simons term ( breaks also parity ) allowed in the lagrangian giving rise to topological corrections in the maxwell equations . the effective lagrangian of the classical dirac field is given by @xmath163 , known as topological massive gauge theory @xcite @xmath164 where @xmath165 describes the fermionic , @xmath166 the gauge field and @xmath167 the interaction part , @xmath168 is the abelian chern - simons term and we have used the conventional definitions @xmath169 and @xmath170 . the gamma matrices fulfill a clifford algebra @xmath171 , where @xmath172 is the minkowski metric . while the following derivation is basis independent , we set @xmath173 , @xmath174 and @xmath175 corresponding to the spin - up block of the bhz hamiltonian . using euler - lagrange formalism , @xmath176 the equations of motion for the gauge field ( maxwell equations ) are derived , @xmath177 where the second equation is the bianchi identity and @xmath178 is the dual field tensor . for a better visualization , we write out the equation of motions in component form , @xmath179 where it should be noted that the curl as well as the magnetic field are pseudoscalar in two space dimensions and @xmath180 . by deriving the continuity equation from eq . ( [ eqa : max1 ] ) and ( [ eqa : max3 ] ) , @xmath181 \epsilon_{\mu\nu}f^{\mu\nu } \label{eqa : continiuitymaxwell}\ ] ] and comparing the result with eq . ( [ eqa : parityanomaly ] ) , we find that @xmath182 since @xmath183 ( see eq . ( [ eqa : parityanomaly ] ) ) , we find that for an arbitrary magnetic field @xmath184 this charge is linked to an either filled or empty @xmath185 ll in the ground state ( noting that @xmath186 corresponds to the carrier density of a single landau level ) , where magnetic flux is tied to the electrons . in an experimental set - up , our theoretical demand of a zero carrier density , @xmath187 , is realized by tuning the carrier density via an external electric gate such that the chemical potential falls into the mass gap . however , we see in the following that instead of the bare charge carrier density @xmath33 , only an effective charge carrier density @xmath34 is accessible by an electric gate . finally , we study consequences of a @xmath151-dependence of the dirac mass term as present in eq . ( [ eq : bhzhamiltonian ] ) and derive results for the `` parity anomaly '' accordingly . by using the semi - classical approach , eq . ( [ eqa : densityresponse ] ) and ( [ eqa : hallresponse ] ) , we find that the expression for the `` parity anomaly '' is renormalized by the bhz parameter @xmath12 and reads @xmath188\epsilon_{\mu\nu\tau}f^{\nu\tau}\label{eqa : fullparity}.\ ] ] by including also the quadratic term ( @xmath189 ) in the lagrangian @xmath190 ( without writing the calculation explicitly ) , we find that the maxwell equations are not changed and our main result , @xmath191 , remains valid . the discussion was presented here for the single spin - up block , however it can be analogously repeated for the spin - down block . the experimental signature is determined by the sum of both spin - blocks . the appendix deals with symmetries of the bhz hamiltonian , in particular with time - reversal ( trs ) and parity symmetry , where the focus is on the latter since we are interested in the origin of the `` parity anomaly '' . each block of the bhz hamiltonian @xmath192 ( eq . ( [ eq : bhzhamiltonian ] ) ) can be interpreted as a massive ( 2 + 1)d dirac equation with a @xmath151-dependent mass term . at first , we discuss symmetries of a single block and in the second part , we focus on the symmetries of the full bhz hamiltonian . for a single block , invariance under pseudospin trs ( in the band index ) implies that an anti - unitary operator @xmath193 exists such that @xcite @xmath194 with @xmath195 where @xmath196 is a unitary matrix and @xmath197 is the operator of complex conjugation . parity symmetry in two space dimensions amounts for e.g. @xmath198 @xcite . invariance under parity symmetry implies then that a unitary operator @xmath199 exists such that @xcite @xmath200 the mass term ( including the newtonian mass ) is odd under both symmetries , e.g. in case of parity symmetry @xmath201 @xmath202 where we have used that @xmath203 . therefore , pseudospin trs as well as parity symmetry is broken by the mass term giving rise to the `` parity anomaly '' in odd spacetime dimensions . however , in the full bhz hamiltonian the anomaly cancels since both spin blocks amount for a different sign in the induced current ( compare with eq . ( [ eq : parityanomaly ] ) ) such that the sum must be zero . the cancellation of the anomalous four - current for the full bhz model can be also understood in the following way : we can define a trs as well as parity operator for the full bhz hamiltonian connecting both spin blocks @xmath204 ( similar to a two flavor ( 2 + 1)d massive dirac equation @xcite ) . such a trs operator exists and is given by @xmath205 while the parity operator is @xmath206 . consequently , there is no `` parity anomaly '' for the pure bhz hamiltonian and the total induced current must vanish . however , in the presence of an exchange or zeeman - like hamiltonian , as given by eq . ( [ eq : zeeman ] ) and ( [ eq : exchange ] ) , parity as well as trs are broken , since e.g. @xmath28 is odd under parity symmetry @xmath207 in an external magnetic field ( applying peierls substitution ) , we find additionally to the zeeman and exchange interaction that another parity breaking term arises which is given by @xmath208 all in all , the exchange interaction , the zeeman as well as the particle - hole asymmetry term break parity and renormalize the effective mass gap in magnetic fields as highlighted in eq . ( [ eq : massterm ] ) . a topological transition is therefore driven by an effective g - factor consisting off all three contributions . starting from a qsh phase , the total four - current remains zero during an adiabatic increase in the effective g - factor as long as the mass gap of one of the spin - blocks is not closed . if one of the mass gaps is closed , the system enters a qah ( requires ferromagnetic exchange interaction without an external magnetic field ) or a qahl ( qah state in magnetic fields ) phase and a non - zero anomalous four - current is observed . finally , one should note that a bia term also violates parity symmetry , where a representation of the bia term given in the dirac basis reads @xcite @xmath209 where @xmath210 is a parameter . however , rashba spin - orbit interaction in lowest order @xcite @xmath211 where @xmath212 is a parameter , does not violate parity . both bia and rashba spin - orbit interaction were found to be negligible ( less than @xmath213 mev ) in recent experiments on symmetric hgte qws @xcite . since the mass gap protects the anomalous four - current and , therefore , the `` parity anomaly '' , such small perturbations can not destroy the `` parity anomaly '' as long as no gap closing occurs .. this appendix should elucidate how effective bhz - parameters ( see eq . ( [ eq : bhzhamiltonian ] ) ) for @xmath214 quantum wells have been numerically calculated . these parameters haven been used in the main text to predict an early topological transition in magnetic fields to a @xmath84 plateau for a @xmath215 thick sample of @xmath108 as depicted in fig . [ fig : landaufan ] . the bulk band structure in case of hgmnte around the @xmath216-point is characterized by the @xmath217 - kane hamiltonian , @xmath218 where @xmath219 is the bare kane hamiltonian , @xmath220 is the zeeman term , @xmath221 describes sp - d exchange interaction between the s-/p - band electrons and localized d - electrons associated to mn atoms and @xmath222 amounts for uniaxial strain . the hamiltonians are discussed in great detail in ref . @xcite , where they are written out explicitly in eq . ( 6 ) , ( 14 ) , ( 20 ) and ( c3 ) , respectively . it is worth noting that mn does not only give rise to an exchange interaction but one must also renormalize kane parameters for a given mn concentration . for small fractions , it is valid to assume that only the band gap parameter @xmath223 is strongly affected @xcite . this affects the critical thickness above which hgmnte becomes a 2d ti , for instance @xmath224 , while in pure hgte the transition should have already emerged at @xmath225 @xcite . the full hamiltonian is then mapped onto a cubic lattice and we assume periodic boundary conditions in x- and y - direction , while in z - direction the interface between hgmnte and cdte is taken into account by having @xmath226-dependent kane parameters @xcite . mapping to a cubic lattice corresponds to making use of a finite difference method ( also known as tight - binding method ) , where e.g. a single matrix element with a z - dependent kane parameter is given by @xmath227\right.\nonumber\\ & & \left . -\gamma{\left(}z-\frac{a}{2}{\right)}\left [ \psi{\left(}z{\right)}-\psi{\left(}z -a{\right)}\right]\right\},\end{aligned}\ ] ] where @xmath228 is an exemplary kane parameter , @xmath229 and in the second step we used a finite difference quotient with grid spacing @xmath230 . the eigenvalue problem of the schrdinger equation can then be solved numerically . using lwdin perturbation theory @xcite up to the second order , one can derive a low energy model which is valid close to the band gap at the @xmath216-point . this model is known as bhz hamiltonian @xcite , where @xmath231 and @xmath232 where @xmath233 and @xmath234 . moreover , the full kane hamiltonian was divided in two parts @xmath235 , where @xmath236 is the hamiltonian at the @xmath216-point and @xmath237 is treated as a perturbation . the indices @xmath238 run over all subbands at @xmath216 forming the basis of the low energy hamiltonian . in case of the bhz hamiltonian this means that @xmath238 compose the four subbands @xmath19 . all other subbands are indexed by @xmath239 and give rise to a renormalization of the bare bhz parameters . finally , we end up with the bhz hamiltonian as it was explicitly given in eq . ( [ eq : bhzhamiltonian ] ) . two exemplary sets of bhz parameters for @xmath240 are given in the caption of fig . [ fig : landaufan ] of the main text . finally , we want to write out the exchange hamiltonian , @xmath241 , as it was given in eq . ( [ eq : exchange ] ) , since it is the exchange coupling which dominates in case of hgmnte the effective g - factor @xcite @xmath242 where @xmath243 where @xmath244 , @xmath245 is the brillouin function , @xmath246 , we assume zero temperature @xmath247 , @xmath248 , @xmath249 and @xmath250 and @xmath251 amounts for s- and p- band character of the @xmath252 subband , respectively . in this appendix , we highlight that the property of the @xmath62 ll being either a solution of the valence band ( occupied in ground state ) or being a solution of the conduction band ( unoccupied in ground state ) is already contained in the landau level spectrum given by eq . ( [ eq : llspectrumchernn ] ) . we focus on the spin - up block while the discussion applies analogously to the spin - down block . therefore , it is possible to make predictions for topological transitions in magnetic field based solely on the expression for landau level energies . a short prove is outlined in the following based on a half - space calculation @xcite . solutions of the schrdinger equation on the half space ( @xmath253 ) , @xmath254 , are obtained imposing the ansatz @xmath255 , where @xmath256 , @xmath257 and @xmath258 are parabolic cylindrical functions . the eigenvalue problem is solved using the recurrence relations @xmath259 after a straightforward calculation , it follows that the spectrum is again given by eq . ( [ eq : llspectrumchernn ] ) , @xmath260 where @xmath52 , @xmath53 and @xmath54 , @xmath55 , @xmath261 with @xmath57 as in the main text . however , with the difference that @xmath262 is replaced by @xmath263 . obviously , there are in general two solutions with @xmath264 , where the respective `` @xmath265 '' - sign in eq . ( [ eqa : llspectrumchernnu ] ) symbolizes that the @xmath62 ll belongs either to the valence or to the conduction band . this is in contrast to the pure bulk calculation presented in the main text , where the eigenvalue of the @xmath62 ll is comprised within a single expression ( see eq . ( [ eq : llspectrumchern0 ] ) ) . this is due to the fact that the boundary conditions have already been applied in the bulk calculation by choosing the hermite polynomials as solutions . a solution of the eigenvalue problem is given by @xmath266 where we have used that @xmath267 and @xmath268 are two independent solutions , @xmath269\end{aligned}\ ] ] and @xmath270 . by applying the boundary conditions , @xmath271 and @xmath272 , we find that only one @xmath264 solution is allowed . depending on whether @xmath273 or @xmath274 , the @xmath62 ll belongs either to the valence or to the conduction band , respectively . such a topological transition in magnetic field is shown in fig . [ figa : halfspace ] . here , the @xmath62 ll changes from the valence ( fig . [ figa : halfspace ] a ) ) to the conduction band ( fig . [ figa : halfspace ] b ) ) and the transition point is marked by a massless dirac fermion , @xmath275 as we accordingly found in appendix a. while the discussion here was only given for the spin - up block , a transition can occur either for the spin up or the spin - down block depending on the sign of the effective g - factor as discussed in the main text . in the bulk limit @xmath276 are indicated by points ( toy parameters ) . in this limit , @xmath172 converges against integer - values . along the topological transition ( for @xmath277 and @xmath278 ) , the @xmath62 ll changes from ( a ) the hole ( red ) to ( b ) the particle branch ( blue ) . to highlight the branch change , the inset in ( b ) shows enlarged the @xmath62 ll . the chemical potential at constant effective carrier density @xmath34 is indicated by the dashed line . during the transition the chern number changes from @xmath66 to @xmath279 . ]	recent developments in solid state physics give a prospect to observe the parity anomaly in ( 2 + 1)d massive dirac systems . here we show , that the quantum anomalous hall ( qah ) state in orbital magnetic fields originates from the dirac mass term and induces an anomalous four - current related to the parity anomaly . this differentiates the qah from the quantum hall ( qh ) state for the experimentally relevant case of an effective constant density ( seen by the gate ) . a direct signature of qah phase in magnetic fields is a long @xmath0 ( @xmath1 ) plateau in cr@xmath2(bi@xmath3sb@xmath4)@xmath5te@xmath6 ( hgmnte quantum wells ) . furthermore , we predict a new transition between the quantum spin hall ( qsh ) and the qah state in magnetic fields , for constant effective carrier density , without magnetic impurities but driven by effective g - factors and particle - hole asymmetry . this transition can be related to the stability of edge states in the dirac mass gap of 2d topological insulators ( tis ) , even in high magnetic fields . _ introduction : _ condensed matter analogs of the ( 2 + 1)d and ( 3 + 1)d dirac equation , i.e. topological insulators ( ti ) and weyl semimetals , opened new directions to study high energy anomalies in the solid state lab . an anomaly in high energy physics is defined as breaking of a classical symmetry during regularization @xcite and , in particular , the parity anomaly is characterized by a broken parity symmetry in a quantized theory . for instance in case of the massless dirac equation in ( 2 + 1)d , coupled to an electromagnetic field , parity symmetry is violated if one insists on gauge - invariance @xcite . in solid state physics on the other hand , `` parity anomaly '' is often understood as generation of an anomalous four - current related to a finite mass term in the dirac equation @xcite . anytime when we relate to the notion of `` parity anomaly '' , we refer to the second definition . at first glance , chiral currents related to the massive dirac equation in ( 2 + 1)d , coupled to an electromagnetic field , appear similar to chiral currents associated with the qh effect ( quantization of hall conductance in a two - dimensional ( 2d ) electron gas in an external out - of - plane magnetic field @xcite ) . however , in case of the `` parity anomaly '' the current is induced by the mass term ( violating parity ) , while the current in the qh phase is induced by magnetic field ( no violation of parity ) @xcite . despite a plethora of suggestions for possible realizations of the `` parity anomaly '' in a condensed matter system , its experimental observation is still outstanding @xcite . recently new topological states of matter have been discovered such as 2d tis characterized by the qsh effect where an odd number of pairs of counterpropagating ( helical ) edge states exist at the boundary @xcite . when 2d tis @xcite or thin films of 3d tis @xcite are doped with magnetic impurities , the gap for one of the spin directions can be closed due to broken time - reversal symmetry ( trs ) and qah effect forms . it is characterized by a single propagating chiral edge state existing even in the absence of magnetic field . when both trs and particle - hole symmetry ( phs ) are broken qsh , qah and qh phases are all classified by a @xmath7-topological invariant @xcite . outstanding questions then arise , whether any of these effects are related to the `` parity anomaly '' and if their experimental distinction is possible in an external magnetic field . answering these questions is the goal of this paper . the qsh phase is described by two copies of massive ( 2 + 1)d dirac hamiltonians @xcite . as we will show in this paper , there is an anomalous four - current for each single block , indexed by @xmath8 , if the chemical potential lies in the mass gap : @xmath9\epsilon_{\mu\nu\tau}f^{\nu\tau},\label{eq : parityanomaly}\ ] ] where @xmath10 is the electromagnetic tensor , while @xmath11 and @xmath12 are the relativistic dirac mass and the effective mass originating from quadratic dispersion ( newtonian mass ) @xcite , respectively . sum and difference of anomalous four - currents , @xmath13 , can be used to distinguish between qh , qah and qsh phases in magnetic field . although the helical edge states of the qsh phase are not anymore protected against backscattering in magnetic fields ( due to broken trs @xcite ) as long as neither @xmath11 nor @xmath12 change their sign as a function of the magnetic field , @xmath14 remains nonzero . we denote this phase qsh - like ( qshl ) phase . similarly , the qah state is characterized by a nonzero @xmath15 as well as @xmath14 in magnetic field and we denote this phase qah - like ( qahl ) . chiral edge states related to the qahl phase form in the mass gap and are therefore distinct from qh edge states which form above the mass gap and have hall currents scaling as the sign of magnetic field @xcite . as pointed out by ma et al . @xcite , the effective carrier density and not the chemical potential can be accessed by a gate electrode in a typical experimental set - up . however , topological transitions in magnetic field have been so far mostly analyzed in terms of constant chemical potential @xcite . therefore , we predict in this paper new types of topological transitions in magnetic field at constant effective carrier density . under this condition , we show that the `` parity anomaly '' allows for a unique distinction of the qahl and qh phases in magnetic fields . in particular , we predict a long hall plateau at @xmath1 ( @xmath16 ) , related to the qahl phase of hgmnte ( @xmath17 ) @xcite . moreover , we predict a new topological qsh to qahl transition , occurring without any magnetization but as a consequence of the particle - hole asymmetry and effective g - factors . this latter transition might be related to the recent observation of stable edge states of 2d tis in the mass gap for high magnetic fields @xcite . _ model : _ let us consider the bhz hamiltonian , i.e. a two flavor-(2 + 1)d massive dirac hamiltonian @xcite , describing 2d tis , @xcite @xmath18 written in the dirac basis , i.e. in the low energy subbands @xmath19 . here , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 and @xmath25 , @xmath11 , @xmath12 and @xmath26 are system parameters . @xmath27 denotes a zeeman hamiltonian and @xmath28 describes exchange interaction between s / p - band electrons with localized spins belonging to the magnetic impurities . for hgmnte @xcite , the exchange coupling is paramagnetic @xcite , while it is ferromagnetic for @xmath17 @xcite . in the following , we demand that the chemical potential is placed in the dirac mass gap at zero magnetic field such that the system is insulating in the bulk . and * ( b ) * a trivial insulator @xmath29 in magnetic fields . @xmath30 state is not shown and stays constant across the transition . the magnetic gap is given by @xmath31 and the mass gap by @xmath32 . for the trivial insulator , both the bare charge @xmath33 and the effective charge @xmath34 are equal to zero ( see main text ) . * ( c ) * schematic view of charge pumping in ground state of qahl insulator . for qahl system an adiabatic increase of the magnetic field @xmath35 causes a circulating electric field . at the mass domain wall , indicated by the colored landscape , chiral edge states form and charges flow from the domain wall to the bulk as a response to the electric field . due to the `` parity anomaly '' , an additional topological term occurs in the chern - simons action of a qahl system such that the net charge / electric field is zero , @xmath36 but @xmath37 . black arrows represent schematically flux quanta attached to the electrons . ] let us first study the bulk landau level ( ll ) spectrum of the spin - up block . we consider an external magnetic field @xmath38 oriented perpendicular to the plane of the 2d electron gas , @xmath39 , with @xmath40 . the orbital effect of the magnetic field is introduced by the peierls substitution @xmath41 @xmath42 , using the landau gauge @xmath43 . the bulk ll spectrum is obtained using ladder operators @xmath44 and @xmath45 fulfilling the usual commutation relation @xmath46=1 $ ] , where @xmath47 is the magnetic length . this can be used to construct an ansatz for the schrdinger equation @xcite , @xmath48 where @xmath49 is the ll index and @xmath50 is a good quantum number . the resulting spectrum is given by @xmath51 where @xmath52 , @xmath53 and @xmath54 , @xmath55 and @xmath56 where @xmath57 is the effective g - factor . since @xmath58 transforms under parity as the usual dirac mass m , it can be interpreted as the renormalized dirac mass in an external magnetic field and it replaces @xmath11 in eq . ( [ eq : parityanomaly ] ) . more details on the symmetry of the bhz hamiltonian are given in appendix b. it is apparent that even for @xmath59 the effective g - factor is nonzero due to a broken phs . for @xmath60 , all lls come in particle / hole pairs reproducing the bulk band gap in the limit @xmath61 . however , the @xmath62 ll lacks a partner @xcite and its respective chiral edge state goes through the mass gap in case of the topologically non - trivial regime @xmath63 . _ trivial versus qahl phases in orbital magnetic fields : _ here , we study the difference between trivial insulating and qahl phases in magnetic fields . the qahl phase is characterized by @xmath64 , while the trivial insulator has @xmath29 . since the two blocks in eq . ( [ eq : bhzhamiltonian ] ) are decoupled , we focus on the spin - up block @xmath65 ( @xmath66 ) and omit the trivial ( @xmath30 ) spin - down block . details of the calculation are given in appendix a. qws with * ( a ) * @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 and with * ( b ) * @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath73 . @xmath80 ( @xmath81 ) indicates a topologically trivial ( non - trivial ) phase @xcite . at @xmath82 , mn impurity spins are randomly oriented and the system is in the qsh phase , @xmath29 . in a magnetic field , the gap for the spin - up block closes at @xmath83 with a transition to a stable @xmath84 qahl plateau . the transition is indicated by crossing of the chemical potential @xmath85 with the spin - up @xmath62 ll . numbers within plot indicate chern numbers . ] we start the discussion from the qah state with @xmath86 , where the ground state is defined by @xmath87 and the minus sign denotes that only valence band states are filled . adiabatically switching on the magnetic field induces an azimuthal electric field , since @xmath88 . based on a semi - classical calculation @xcite , it is easy to show that an anomalous four - current is induced , @xmath89\epsilon_{\mu\nu\tau}f^{\nu\tau}. \label{eq : anomalouslin}\ ] ] this means that the electric field pump current @xmath90 from the left ( right ) side of the sample into the bulk , @xmath91 ( @xmath92 ) , resulting in a carrier density of @xmath93 with @xmath94 . this corresponds to filling of the @xmath62 ll and signs of currents and charges are determined by the sign of the mass , @xmath95 as schematically indicated in fig . [ fig : qhvsqah]c . since during the process of switching of @xmath35 , the topological dirac mass does not close , the question arises : why does the ground state of qahl phase change from @xmath96 in a zero magnetic field to a finite value for a non - zero @xmath35 ? does this not break the adiabatic assumption for switching on magnetic field as well as the constraint of constant carrier density in experiments @xcite ? the solution comes from the fact that qah and qahl phases break parity symmetry due to the topological mass , @xmath58 ( see appendix b ) , which enforces adding a topological chern - simons term in the gauge field lagrangian , @xmath97 where @xmath98 /h$ ] is the chern - simons coupling constant , which modifies the maxwell equation , @xmath99 the term originating from the chern - simons action generates a flux @xmath100 which compensates charge such that the full effective charge @xmath34 remains zero ( @xmath37 ) . when the chemical potential is located within the mass gap at @xmath101 with @xmath102 , it remains in the mass gap for non - zero @xmath35 . using a half - space calculation , we additionally prove in appendix d that the @xmath62 ll makes a transition from the trivial to qahl insulator at @xmath103 @xcite . even so the discussion was given for a qahl state with @xmath102 , it applies as well for @xmath104 . in comparison , a trivial insulator does not break parity symmetry and therefore no intrinsic chern - simons term is allowed in the gauge field lagrangian @xcite . this is shown in fig . [ fig : qhvsqah]b , where @xmath105 . this means that qahl phase has a non - zero chern number but @xmath106 and therefore it has to be counted differently from qh lls forming above the magnetic gap . therefore , the `` parity anomaly '' distinguishes unambiguously qahl effect from a trivial insulator in magnetic fields . _ stable qahl plateau for @xmath107 : _ let us now come back to the full bhz hamiltonian and compare ll fans for topological trivial and non - trivial insulators . the discussion is given for realistic parameters for hgmnte @xcite . technical details concerning these parameters are given in appendix c. @xmath108 with @xmath67 is a topologically trivial insulator ( @xmath109 ) , the chemical potential for an effective charge @xmath110 lies in the mass gap ( fig . [ fig : landaufan ] a ) ) and it remains to be a trivial insulator even for large magnetic fields . @xmath108 with @xmath74 ( fig . [ fig : landaufan]b ) forms qsh phase with @xmath111 and @xmath112 . for finite @xmath35 , mn becomes polarized and the mass gap for the spin - up block closes at @xmath113 signalizing formation of a massless dirac fermion . for @xmath114 , it is easy to prove ( appendix a ) that @xmath115 while @xmath116 . across this transition , the system goes into a qahl @xmath117 phase . in the literature , topological transitions are often discussed based on the assumption of a constant chemical potential and the qsh to qah transition occurs at zero orbital field @xcite . our transition differs from the above mentioned in two aspects : 1 ) we assume a constant effective density 2 ) the topological transition happens in finite magnetic fields . therefore , we claim that even at non - zero magnetic field , one can define qahl phase with @xmath118 different from zero . in conclusion , we predict a stable @xmath119 plateau with a hall conductance @xmath120 ( see fig . [ fig : qshltotrivial ] ) for hgmnte above critical qw thickness . this prediction is exactly along the lines of recent experiments on hgmnte @xcite , where one needs non - zero magnetic field to polarize mn and observes a stable @xmath104 plateau . similarly , recent experiments on @xmath17 show a stable @xmath121 plateau up to @xmath122 @xcite . we believe that indeed these stable plateaus are direct signatures of the `` parity anomaly '' in these systems and the sign of g - factors determines if the stable plateau occurs for @xmath123 or @xmath124 . furthermore , after crossing of the two @xmath62 lls at @xmath125 @xcite , the adiabatic condition is possibly not anymore valid . we leave the discussion of this point to another paper , assuming for the moment being that the transition to the trivial phase occurs @xcite . the experimental signature in case of hgmnte is therefore reentrant behavior of the hall conductance starting from @xmath126 in the qsh phase , changing to @xmath127 in the qahl state , and reentering @xmath128 plateau for @xmath129 ( see fig . [ fig : qshltotrivial ] ) . we have omitted off - diagonal terms connecting the two dirac hamiltonians such as rashba spin - orbit interaction terms @xcite or bulk - inversion asymmetry terms @xcite . they should be small in comparison to diagonal terms and should only renormalize the effective mass gap slightly @xcite . topological transitions in magnetic fields are induced by mass gaps closing and as long as these off - diagonal terms do not close the gap , we expect that only the critical magnetic field might deviate from our prediction . ( see also appendix b ) . . left axis corresponds to the non - local four - terminal hall conductance , @xmath130 ( as for example in @xcite ) , while the right axis to the local two - terminal longitudinal conductance , @xmath131 ( as for example in @xcite ) . to highlight transitions we subdivided the plot in three areas . in the insets the blue and red lines show schematically the @xmath62 lls and the black dashed line shows the position of the chemical potential in the ground state . @xmath128 for @xmath132 for qshl effect . the transition to qahl phase with a @xmath84 occurs at @xmath133 by virtue of the `` parity anomaly '' . at the crossing point of the two @xmath62 ll , @xmath125 , the system might become trivial ( see main text ) . the @xmath134 in the qshl phase changes to @xmath135 at @xmath133 ( qahl phase ) , and possibly to zero at @xmath125 . ] _ qshl to qahl transition in non - magnetic tis : _ another hallmark of the `` parity anomaly '' is a transition from a qshl into a qahl phase in magnetic fields even without magnetic impurities . as pointed out before , for hgmnte qws the exchange interaction is paramagnetic and , therefore , magnetic field dependent @xcite . values of the critical field @xmath133 must be therefore found numerically . however , if we omit exchange coupling ( @xmath136 ) , one obtains an analytical expression for a transition from the qsh phase to qahl phase . closing of the gap for a spin - up block corresponds to @xmath113 with @xmath137 , where @xmath138 is the effective g - factor without exchange interaction . interestingly this transition can occur without any magnetization or magnetic impurities but purely from the particle - hole asymmetry and effective g - factors terms which act in magnetic field as the mass connected with the `` parity anomaly '' . we expect that in the qahl phase , the hall conductance of the edge states should be more precisely quantized in comparison to the qshl phase and survive even in large magnetic fields ( due to lack of backscattering ) . in particular , we predict a transition from qshl phase , characterized by a local two - terminal conductance of @xmath139 , to a qahl phase at @xmath140 , characterized by @xmath141 , and finally to the trivial phase at @xmath125 ( see green dashed line in fig . [ fig : qshltotrivial ] ) . therefore , our prediction of surviving topological edge states in high magnetic fields could be related to the recent observation in ref . @xcite . summarizing , we have studied topological transitions in magnetic fields for a two - flavour ( 2 + 1)d massive dirac hamiltonian . we have studied the experimentally relevant case of an effective constant carrier density with the chemical potential located within the mass gap . we have demonstrated that the qahl phase ( qah effect in orbital magnetic fields ) can be distinguished from the qh phase since the former one is a direct consequence of the `` parity anomaly '' for massive dirac systems . we have shown that a long @xmath124 and @xmath142 plateaus in hgmnte and @xmath17 , respectively , are direct signatures of qahl phase . moreover appendix a , we have predicted a new topological transition from the qsh phase to the qahl phase without magnetization or magnetic impurities , which can explain the stability of edge states in recent experiments @xcite . _ acknowledgments . _ we thank f. wilczek , b. a. bernevig , c. brne , l. w. molenkamp , c. kleiner , j. s. hofmann , j. erdmenger , c. morais smith and w. beugeling for useful discussions . we acknowledge financial support from the dfg via sfb 1170 `` tocotronics '' , and the enb graduate school on topological insulators . 43ifxundefined [ 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 _ _ , edited by ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop _ _ , edited by ( , ) @noop * * , ( ) [ appendix ]
in the main text of this paper , we have obtained edge - state spectra for finite - size disk systems with different boson numbers , whose ground states locate at various angular momenta . also the periodicity of the energy spectra under flux insertion depends on particle numbers , which is @xmath83 when boson numbers satisfy @xmath80 with @xmath81 and @xmath82 as coprime integers . we can indeed heuristically understand these numerical results in view of the generalized pauli principle which rules the boson occupancy in single - particle orbitals with particular angular momenta , and considering that many - body states are built from single - particle orbitals with particular flux periodicity . sites and the trap potential @xmath85 . ( a ) single - particle energy spectrum . ( b ) zoom in on the lowest energy levels of ( a ) . red circles represents the hardcore bosons occupied in single - particle orbitals ( the number labelled below each low energy level is the angular momentum quantum number ) , the configuration is ruled by the generalized pauli principle . the ground state locates at the angular momentum @xmath86 , with subsequent low - energy excited states lying in incremental momentum sector @xmath87 where @xmath88 denotes @xmath88th lowest energy state . ] let us first look at the single - particle orbitals of these finite disk systems , where we take kagom lattice with 72 sites ( corresponding to the system studied in fig . 3 of the main text ) as an example . parameters for the kagom lattice hamiltonian are kept the same as in the main text . the energy spectrum for this tiny system is illustrated in fig . [ kg_1b_edge ] , showing the single - particle ground state locating at angular momentum @xmath89 , with subsequent low - energy excited states lying in incremental momentum sector @xmath87 where @xmath88 denotes @xmath88th lowest energy state . considering @xmath16 rotational symmetry of the system , angular momentum @xmath90 could also be written as @xmath91 , as is shown in the energy spectrum . by inserting a central flux into this system , we are able to see that a series number of low energy states , which are 18 states in the case of kagom lattice with @xmath92 sites ( corresponding to the system shown in fig . 4 of the main text , where we have studied the flux insertion ) , evolve into each other without mixing with other higher energy states ; see fig . [ 1b_flux ] . when the flux changes its value by @xmath69 , we can clearly see that a single - particle state with angular momentum @xmath90 evolves into the state with angular momentum @xmath93 . we will see in the subsequent part how to infer the flux periodicity of many - body states which are built from single - particle orbitals with particular flux periodicity . sites and the trap potential @xmath23 . a single - particle state with the angular momentum @xmath90 evolves into the state with the angular momentum @xmath93 . different colors represents the energy states from different angular momentum sectors . ] when a few number of hardcore bosons are loaded into the finite disk system , they occupy these low - energy states with the restriction that no more than one particle occupy two consecutive orbitals , which is stated in the generalized pauli principle for the @xmath15 fci / fqah phases ( refs . [ 8,42 ] of the main text ) . based on this simple picture , both questions mentioned at the beginning can be well explained without difficulty . following above , the many - body ground state in a @xmath15 fci / fqah phase can be indexed as the `` root configuration '' @xmath94 ( refs . [ 42 ] of the main text ) , where @xmath95 is the particle number and @xmath96 denotes the boson number occupying the @xmath88th lowest - energy state . the total angular momentum @xmath97 can be written as @xmath98 . so @xmath99 , @xmath100 , @xmath101 , @xmath102 , @xmath103 , @xmath104 , which is in exact accordance with our numerical results . we confirmed this gs angular momentum analysis applies to the situation of kagom lattice ( e.g. the 72-site cases in fig . 3 of the main text ) . in addition , this analysis also applies to the honeycomb - lattice haldane model ( e.g. the 96-site cases in fig . 2 of the main text ) . we now turn to the situation where a central flux is inserted into the finite - size disk system . flux insertion is expected to shift the angular momenta of the many - body states as well as the single - particle ones . when the flux changes its value by @xmath69 , each single - particle state moves to the next angular momentum sector . we should keep in mind that many - body states are built from single - particle orbitals with this particular flux periodicity . for a disk system with @xmath48 bosons , this leads to @xmath105 . using @xmath80 , we obtain @xmath106 , with @xmath81 and @xmath82 as coprime integers . considering the @xmath16 rotational symmetry of the system , we finally arrive at the flux periodicity of many - body states ( which is stated in the main text as an empirical fact ) as shown below , @xmath107 hardcore bosons ruled by the generalized pauli principle . the short blue lines represent the single - particle orbitals with the angular momentum quantum numbers labelled below , and the red circles denotes the occupancy of hardcore bosons . the total angular momentum quantum numbers are labelled at the left as `` @xmath108 '' etc . ] in the main text of this paper , we numerically find that edge excitations of finite disk systems in fractional chern insulators exhibit a quasi - degeneracy sequence , which turns out to be quite coincident with that predicted by the chiral luttinger liquid theory . in this section , we will show that the degeneracy sequence of these edge excitation spectra can also be well predicted using the generalized pauli principle . the key idea is to count the number of ways to put hardcore bosons in single - particle orbitals for a certain total angular momentum @xmath109 , which should be equal to the degeneracy level in sector @xmath109 ; the allowed configurations ( ref . [ 42 ] of the main text ) should be ruled by the generalized pauli principle . for a finite disk system , we expect a good match in numerical results when the particle number is small , where finite - size effect is not so obvious . without loss of generality , we discuss a disk system with @xmath58 hardcore bosons . we mentioned in early sections that the root configuration of ground state should be @xmath110 , where @xmath111 denotes subsequent 0 s in the case of @xmath58 bosons , and the total angular momentum should be @xmath112 ( or 3 after @xmath113 ) . since this is the only configuration satisfying @xmath108 , degeneracy level @xmath114 in this sector should be @xmath115 . in the @xmath116 sector , the only allowed configuration is @xmath117 , where @xmath118 , so @xmath119 . in the @xmath120 sector , the allowed configurations are @xmath121 , where @xmath122 , and @xmath123 , where @xmath124 , so @xmath125 . in this manner , we can deduce the degeneracy level at each angular momentum for a given number of particles . an illustration is shown in fig . [ degeneracy ] , which is @xmath126 for systems with @xmath50 , exactly coincident with what we have obtained from numerical calculations ! to conclude , we are able to heuristically interpret some important features of edge excitations even in a rather tiny system based on the generalized pauli principle . it is also interesting to notice that this principle applies generally to different lattice models and to different system sizes , for small boson numbers that the finite - size effect is ignorable .	recent theoretical works have demonstrated the realization of fractional quantum anomalous hall states ( also called fractional chern insulators ) in topological flat band lattice models without an external magnetic field . such newly proposed lattice systems play a vital role to obtain a large class of fractional topological phases . here we report the exact numerical studies of edge excitations for such systems in a disk geometry loaded with hard - core bosons , which will serve as a more viable experimental probe for such topologically ordered states . we find convincing numerical evidence of a series of edge excitations characterized by the chiral luttinger liquid theory for the bosonic fractional chern insulators in both the honeycomb disk haldane model and the kagom - lattice disk model . we further verify these current - carrying chiral edge states by inserting a central flux to test their compressibility . _ introduction. _ one of the most essential and fascinating topics in condensed matter physics is to explore and classify the various states of matter , among which the integer quantum hall effect ( iqhe ) @xcite and the fractional quantum hall effect ( fqhe ) @xcite have long been the major focus . it is well known that , at fractional fillings , the fqhe will emerge when interacting particles move in landau levels ( lls ) caused by an external uniform magnetic field . on the other hand , great interest has been aroused to realize both the iqhe and fqhe in lattice models in the absence of an external magnetic field . an initial theoretical attempt was made by haldane @xcite , who proposed a prototype lattice model to achieve the iqhe by introducing two non - trivial topological bands with chern numbers @xmath0 @xcite . haldane s model demonstrates that the iqhe can also be attained without lls , i.e. defines the quantum anomalous hall ( qah ) states . the lattice version of the fqhe without lls , however , comes much latter because of its intriguing strongly correlated nature . the recent proposal of topological flat bands ( tfbs ) fulfils the basic ingredients to explore the intriguing fractionalization phenomenon without lls . tfbs @xcite belong to a class of band structures with at least one nearly flat band with non - zero chern number . this may be viewed as the lattice counterpart of the continuum lls . several systematic numerical works have been done to explore the correlation phenomenon within tfb models , and abelian @xcite and non - abelian @xcite fqhe states without lls have already been well established numerically . this intriguing fractionalization effect in tfbs without an external magnetic field , defines a new class of fractional topological phases , i.e. fractional quantum anomalous hall ( fqah ) states , now also known as fractional chern insulators ( fcis ) . some new approaches , e.g. the wannier - basis model wave functions and pseudo - potentials @xcite , the projected density operator algebra @xcite , the parton wave - function constructions @xcite , and also the adiabatic continuity paths @xcite , have been proposed to further understand these fci / fqah states in tfbs . there are various other proposals of tfb models and material realization schemes @xcite . very recent systematic numerical studies have also found exotic fci / fqah states in tfbs with higher chern numbers ( @xmath1 ) which do not have the direct continuum analogy in lls @xcite , and also the hierarchy fci / fqah states @xcite . rotational symmetry . various disk sizes are indicated by the circles and the labelled numbers . ] rotational symmetry . various disk sizes are indicated by the circles and the labelled numbers . ] although previous studies have firmly established various properties of a large class of fci / fqah phases , we are here concerned with the less studied edge excitations since they in principle should open another window to reveal the bulk topological order @xcite . edge excitations might also provide a more viable experimental probe especially considering about the possible future realizations of fci / fqah phases in optical lattice systems @xcite . some very promising and detailed experimental schemes have been proposed for realizing the tfb models and the fci / fqah phases , e.g. implementing dipolar spin systems with ultracold polar molecules trapped in a deep optical lattice driven by spatially modulated electromagnetic fields @xcite . the chiral edge modes of such topological phases are expected to be directly visualized in optical - lattice - based experiments @xcite . a recent work has studied edge excitations of the bosonic fqhe in lls generated by an artificial uniform flux in optical lattices @xcite , where finite - size disks and cylinders with trap potentials have been considered , and the edge spectrum is clearly observed . also , related knowledge of fqhe edge states also exists in the study of orbital entanglement spectra @xcite . in the present work , we investigate edge excitations of the bosonic fci / fqah phases @xcite of tfb models in a disk geometry . through extensive systematic numerical exact diagonalization ( ed ) studies , we demonstrate clear evidence of edge excitation spectra of bosonic fci / fqah phases . these edge excitation spectra are quite coincident with the chiral luttinger liquid theory @xcite . to further check the compressibility of these edge states , we insert a central flux into the disk system and indeed verify these current - carrying chiral edge states upon tuning the flux strength . _ formulation. _ we first look into the haldane model @xcite on a honeycomb - lattice disk , which is loaded with interacting hard - core bosons @xcite : @xmath2\\ & -&t\sum_{\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle } \left[b^{\dagger}_{\mathbf{r}^{\prime}}b_{\mathbf{r}}+\textrm{h.c.}\right ] -t^{\prime\prime}\sum_{\langle\langle\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle\rangle\rangle } \left[b^{\dagger}_{\mathbf{r}^{\prime}}b_{\mathbf{r}}+\textrm{h.c.}\right]\nonumber \label{e.1}\end{aligned}\ ] ] where @xmath3 creates a hard - core boson at site @xmath4 , @xmath5 , @xmath6 and @xmath7 denote the nn , the nnn and the next - next - nearest - neighbor ( nnnn ) pairs of sites , respectively . we adopt the previous parameters @xcite to achieve a lowest tfb with a flatness ratio of about 50 ( the ratio of the band gap over bandwidth ) : @xmath8 , @xmath9 , @xmath10 and @xmath11 . we also consider a kagom lattice tfb model @xcite which is also loaded with hard - core bosons , and the model hamiltonian has the form : @xmath12\nonumber\\ & -&t^{\prime}\sum_{\langle\langle\mathbf{r}\mathbf{r}^{\prime}\rangle\rangle } \left[b^{\dagger}_{\mathbf{r}^{\prime}}b_{\mathbf{r}}+\textrm{h.c.}\right]%\nonumber\\ \label{e.2}\end{aligned}\ ] ] we choose the previous parameters for this kagom - lattice model @xcite : @xmath8 , @xmath13 , @xmath14 , which leads to a lowest tfb with the flatness ratio of about 20 . convincing numerical evidence of the @xmath15 bosonic fci / fqah states at the @xmath15 hard - core boson filling of a lowest tfb on a torus geometry has been well established previously @xcite , with the formation of a quasi - degenerate ground - state ( gs ) manifold , characteristic gs momenta , a robust bulk excitation spectrum gap , as well as a fractional quantized chern number for each gs . the disk geometries for both lattice models are illustrated in fig . [ disk ] , which shows the @xmath16 rotational symmetry . an additional trap potential is required on the finite - size disk systems to confine the fci / fqah droplet , outside which the edge modes are able to propagate around the disk . here we choose the conventional harmonic trap , of the form @xmath17 with @xmath18 as the potential strength @xcite ( with the nn hopping @xmath19 as the energy unit ) , @xmath20 as the radius from the disk center ( with the half lattice constant @xmath21 as the length unit ) , and @xmath22 as the boson number operator . sites and trap potential @xmath23 . numbers labelled upon low energy levels in each sector show the quasi - degeneracy of low edge excitations , which is also the sequence derived from the chiral luttinger liquid theory . ] _ edge excitations._we have first considered the honeycomb disk model , and have observed clear edge excitation spectra for a relative broad range of the trap potential strength @xmath18 and various honeycomb disk sizes with @xmath24 sites . some representative results are shown in fig . [ hc_edge ] . as the system owns the @xmath16 rotational symmetry , each energy states can be classified with a quantum number of angular momentum @xmath25 . these edge - excitation quasi - degeneracies of a finite lattice system with various boson numbers are expected to match with that predicted by the chiral luttinger liquid theory for the lowest few sectors @xcite . .occupation of chiral edge bosons in @xmath26 ... orbitals : \{@xmath27 , ... } . @xmath28 is the degeneracy of the edge excitations in a sector . @xmath29 , which is the shifted total angular momentum relative to the ground state angular momentum . the right column shows the degeneracy in systems with finite boson numbers , e.g. , for the `` 4b '' column ( @xmath30 ) , the predicted degeneracy is @xmath31 in @xmath32 angular momentum sectors . the partial degeneracy sequences are mostly observed ( except the underlined @xmath33 , @xmath34 and @xmath35 ) in our ed results . [ cols="^,<,<,^,^,^,^,^",options="header " , ] according to the hydrodynamical approach @xcite , low energy edge excitations of a laughlin - state fqhe droplet at filling @xmath36 are generated by the kac - moody algebra , and form a chiral luttinger liquid with the effective hamiltonian : @xmath37 with the angular momentum @xmath38 along the edge and the fourier - transformed one - dimensional density operator @xmath39 , @xmath40=\frac{\nu}{2\pi}k\delta_{k+k'}$ ] , @xmath41 and @xmath42 are chiral boson / phonon operators . based on this theory , chiral edge bosons occupy in @xmath43 ( unit @xmath44 and edge length @xmath45 ) angular momentum orbitals with the occupation numbers \{@xmath27 , ... } . we denote @xmath46 as the shifted total angular momentum relative to the gs angular momentum . for a given @xmath47 , it is easy to count the edge - state degeneracies by numerating the allowed occupation configurations . in the thermodynamics limit with infinite boson numbers , the degeneracy sequence should be 1,1,2,3,5,7,11,15,22 , ... for our systems with only small boson number @xmath48 s , edge excitations made up of @xmath49 single modes are expected . for three bosons ( @xmath50 ) , after discarding some configurations in the middle column of table [ luttinger ] with @xmath51 , the predicted degeneracy should be 1,1,2,3,4,5,7 , ... , which is in good accordance with the observed quasi - degeneracy in our ed results [ fig . [ hc_edge](b ) ] . for four bosons ( @xmath30 ) , after discarding some configurations with @xmath52 , the predicted degeneracy should be 1,1,2,3,5,6 , ... , which is also in good accordance with the observed quasi - degeneracy in our ed results [ fig . [ hc_edge](c ) ] . the right column in table [ luttinger ] lists the partial degeneracy sequences which are mostly observed in our ed results . sites and trap potential @xmath23 . numbers labelled upon low energy levels in each sector show the quasi - degeneracy of low edge excitations , which is also the sequence derived from the chiral luttinger liquid theory . ] now we turn to investigate the kagom disk model . we studied the finite system also in several different sizes , with @xmath53 sites . edge spectra from a finite kagom disk with @xmath54 sites are shown in fig . [ kg_edge ] , which shows even clearer edge excitation spectra than that obtained from @xmath55 honeycomb disk . it is quite exciting to observe that these edge excitations are independent of lattice geometry even in such small systems . for an example with six bosons ( @xmath56 ) , after discarding some states with @xmath57 , the predicted degeneracy should be 1,1,2,3,5,7 , ... which is also in exact accordance with the observed quasi - degeneracy in our ed results [ fig . [ kg_edge](d ) ] . it is also observed that with the larger disk sizes and smaller boson numbers , the number of matched momentum sectors increases , and thus displays less severe finite size effects . for both disk models , we have noticed that the gss locate at various angular momenta for different boson numbers . the gs angular momenta can be heuristically analyzed from the generalized pauli principle @xcite , which states that no more than one particle occupy two consecutive orbitals for @xmath15 fci / fqah phases , and proves to be a simple yet helpful tool in understanding numerical results @xcite . sites , trap potential @xmath23 , and various boson numbers @xmath48 s . ( a ) for @xmath58 , low energy states in @xmath59 sectors ( red , green respectively ) evolve into each other . ( b ) for @xmath60 , low energy states in @xmath61 sectors ( red , green , blue respectively ) evolve into each other . ] _ central flux._as a next step , we insert a flux into the center of disk geometry ( as shown in fig . [ disk ] ) and tune the flux strength to test the compressibility of edge excitations . consider the kagom disk with @xmath62 sites for an example ; the trap potential will always be fixed as @xmath23 . for the system with @xmath50 bosons , low energy levels @xmath63 in @xmath59 sectors ( or @xmath64 and @xmath65 sectors ) evolve into each other as shown in fig . [ flux](a ) : the low energy levels in the @xmath66 sector ( red ) and those in the @xmath67 sector ( green ) are interchanged when the central flux @xmath68 changes its value by @xmath69 , i.e. @xmath70 ( where @xmath71 ) . these energy spectra @xmath63 finally return to themselves after two periods of the central flux , i.e. @xmath72 . it is obvious to observe the current - carrying nature , i.e. the compressibility of edge states since all of them evolve into the higher energy spectra without any gap . for the system with @xmath30 bosons , low energy levels @xmath63 in @xmath61 sectors ( or @xmath73 sectors ) evolve into each other as is shown in fig . [ flux](b ) : the low energy levels in the @xmath66 sector ( red ) , the @xmath74 sector ( green ) and the @xmath75 sector ( blue ) are cyclically interchanged when the central flux @xmath68 changes its value by @xmath69 or @xmath76 , i.e. @xmath77 ( where @xmath78 ) . these states finally return to themselves after three periods of the central flux , i.e. @xmath79 . for our disk systems with the @xmath16 rotational symmetry , when boson numbers satisfy @xmath80 with @xmath81 and @xmath82 as coprime integers , there is a generic periodicity of energy spectra : @xmath83 . such periodicity can be heuristically understood by considering that the many - body states are built from single - particle orbitals with particular flux periodicity @xcite . _ summary and discussion._we considered two representative tfb lattice models in disk geometry to explore edge excitations of fci / fqah states . with a confining harmonic trap potential , very clear edge excitation spectra are observed with the quasi - degeneracy counting rule satisfying the chiral luttinger liquid theory at least for the lowest six sectors . by inserting and tuning a central flux to both systems , we are able to examine the compressibility of these current - carrying chiral edge states . our work here focuses on the bosonic fci / fqah states in tfbs with the chern number @xmath84 , while it is very interesting to explore the exotic edge excitations in high - chern - number fci / fqah phases @xcite and also those in the hierarchy fci / fqah phases @xcite in future works . we also expect that edge excitation spectra provide another window to reveal the bulk topological order , as well as a more viable experimental probe to future fci / fqah experimental systems . we acknowledge prof . d. n. sheng for helpful discussions and previous collaborations . this work is supported by the nsfc of china grants no . 10904130 , no . 11374265 ( y.f.w . ) and no . 11274276 ( c.d.g . ) , and the state key program for basic researches of china grant no . 2009cb929504 ( c.d.g . ) . _ note added._after the completion of the present work , we became aware of a very recent related preprint @xcite which reports the orbital entanglement spectrum study of the bulk - edge correspondence and edge states in fcis . 99 k. v. klitzing , g. dorda , and m. pepper , phys . rev . lett . * 45 * , 494 ( 1980 ) . d. c. tsui , h. l. stormer , and a. c. gossard , phys . rev . lett . * 48 * , 1559 ( 1982 ) . f. d. m. haldane , phys . rev . lett . * 61 * , 2015 ( 1988 ) . d. j. thouless , m. kohmoto , m. p. nightingale , and m. den nijs , phys . rev . lett . * 49 * , 405 ( 1982 ) . e. tang , j. w. mei , and x. g. wen , phys . rev . lett . * 106 * , 236802 ( 2011 ) ; k. sun , z. c. gu , h. katsura , and s. das sarma , phys . rev . lett . * 106 * , 236803 ( 2011 ) ; t. neupert , l. santos , c. chamon , and c. mudry , phys . rev . lett . * 106 * , 236804 ( 2011 ) . d. n. sheng , z. c. gu , k. sun , and l. sheng , nature commun . * 2 * , 389 ( 2011 ) . y. f. wang , z. c. gu , c. d. gong , and d. n. sheng , phys . rev . lett . * 107 * , 146803 ( 2011 ) . n. regnault and b. a. bernevig , phys . rev . x * 1 * , 021014 ( 2011 ) . y. f. wang , h. yao , z. c. gu , c. d. gong , and d. n. sheng , phys . rev . lett . * 108 * , 126805 ( 2012 ) . b. a. bernevig and n. regnault , phys . rev . b * 85 * , 075128 ( 2012 ) . y. l. wu , b. a. bernevig , and n. regnault , phys . rev . b * 85 * , 075116 ( 2012 ) . x. l. qi , phys . rev . lett . * 107 * , 126803 ( 2011 ) ; m. barkeshli and x. l. qi , phys . rev . x * 2 * , 031013 ( 2012 ) ; y. l. wu , n. regnault , and b. a. bernevig , phys . rev . lett . * 110 * , 106802 ( 2013 ) ; c. h. lee , r. thomale , and x. l. qi , phys . rev . b * 88 * , 035101 ( 2013 ) ; c. m. jian and x. l. qi , arxiv:1303.1787 . s. a. parameswaran , r. roy , and s. l. sondhi , phys . rev . b * 85 * , 241308 ( 2012 ) ; m. o. goerbig , eur . phys . j. b * 85 * , 15 ( 2012 ) . g. murthy and r. shankar , arxiv:1108.5501 ; g. murthy and r. shankar , phys . rev . b * 86 * , 195146 ( 2012 ) . y. m. lu and y. ran , phys . rev . b * 85 * , 165134 ( 2012);j . mcgreevy , b. swingle , and k. a. tran , phys . rev . b * 85 * , 125105 ( 2012 ) ; a. vaezi , arxiv:1105.0406 ; y. zhang and a. vishwanath , phys . rev . b * 87 * , 161113(r ) ( 2013 ) . t. scaffidi and g. mller , phys . rev . lett . * 109 * , 246805 ( 2012 ) ; y. h. wu , j. k. jain , and k. sun , phys . rev . b * 86 * , 165129 ( 2012 ) ; z. liu and e. j. bergholtz , phys . rev . b * 87 * , 035306 ( 2013 ) . x. hu , m. kargarian , and g. a. fiete , phys . rev . b * 84 * , 155116 ( 2011 ) . d. xiao , w. zhu , y. ran , n. nagaosa , and s. okamoto , nature commun . * 2 * , 596 ( 2011 ) . j. w. f. venderbos , m. daghofer , and j. van den brink , phys . rev . lett . * 107 * , 116401 ( 2011 ) . p. ghaemi , j. cayssol , d. n. sheng , and a. vishwanath , phys . rev . lett . * 108 * , 266801 ( 2012 ) . f. wang and y. ran , phys . rev . b * 84 * , 241103(r ) ( 2011 ) . j. w. f. venderbos , s. kourtis , j. van den brink , and m. daghofer , phys . rev . lett . * 108 * , 126405 ( 2012 ) . c. weeks , and m. franz , phys . rev . b * 85 * , 041104(r ) ( 2012 ) . r. liu , w. c. chen , y. f. wang , and c. d gong , j. phys . : condens . matter * 24 * , 305602 ( 2012 ) . w. c. chen , r. liu , y. f. wang , and c. d. gong , phys . rev . b * 86 * , 085311 ( 2012 ) . m. trescher and e. j. bergholtz , phys . rev . b * 86 * , 241111(r ) ( 2012 ) . s. yang , z. c. gu , k. sun , and s. das sarma , phys . rev . b * 86 * , 241112(r ) ( 2012 ) . a. g. grushin , t. neupert , c. chamon , and c. mudry , phys . rev . b * 86 * , 205125 ( 2012 ) . n.y . yao , c.r . laumann , a.v . gorshkov , s.d . bennett , e. demler , p. zoller , and m.d . lukin , phys . rev . lett . * 109 * , 266804 ( 2012 ) ; n. y. yao , a. v. gorshkov , c. r. laumann , a. m. luchli , j. ye , m. d. lukin , phys . rev . lett . * 110 * , 185302 ( 2013 ) . v. yannopapas , new j. phys . * 14 * , 113017 ( 2012 ) . y. zhang and c. zhang , phys . rev . a * 87 * , 023611 ( 2013 ) . z. liu , z. f. wang , j. w. mei , y. s. wu , and f. liu , phys . rev . lett . * 110 * , 106804 ( 2013 ) t. shi and j. i. cirac , phys . rev . a * 87 * , 013606 ( 2013 ) . n. r. cooper and j. dalibard , phys . rev . lett . * 110 * , 185301 ( 2013 ) . y. f. wang , h. yao , c. d. gong , and d. n. sheng , phys . rev . b * 86 * , 201101(r ) ( 2012 ) . z. liu , e. j. bergholtz , h. fan , and a. m. luchli , phys . rev . lett . * 109 * , 186805 ( 2012 ) . a. sterdyniak , c. repellin , b. a. bernevig , and n. regnault , phys . rev . b * 87 * , 205137 ( 2013 ) . t. liu , c. repellin , b. a. bernevig , and n. regnault , phys . rev . b * 87 * , 205136 ( 2013 ) ; a. m. luchli , z. liu , e. j. bergholtz , and r. moessner , phys . rev . lett . * 111 * , 126802 ( 2013 ) . x. g. wen , adv . phys . * 44 * , 405 ( 1995 ) . t. d. stanescu , v. galitski , j. y. vaishnav , c. w. clark , and s. das sarma , phys . rev . a * 79 * , 053639 ( 2009 ) ; n. goldman , j. beugnon , and f. gerbier , phys . rev . lett . * 108 * , 255303 ( 2012 ) ; n. goldman , j. dalibard , a. dauphin , f. gerbier , m. lewenstein , p. zoller , i. b. spielman , pnas * 110 * , 6736 ( 2013 ) . j. a. kjll and j. e. moore , phys . rev . b * 85 * , 235137 ( 2012 ) . a. m. luchli , e. j. bergholtz , j. suorsa , and m. haque , phys . rev . lett . * 104 * , 156404 ( 2010 ) ; z. liu , e. j. bergholtz , h. fan , and a. m. luchli , phys . rev . b * 85 * , 045119 ( 2012 ) . b. a. bernevig and f. d. m. haldane , phys . rev . lett . * 100 * , 246802 ( 2008 ) ; e. j. bergholtz , j. kailasvuori , e. wikberg , t. h. hansson , and a. karlhede , phys . rev . b * 74 * , 081308(r ) ( 2006 ) ; a. seidel and d. h. lee , phys . rev . lett . * 97 * , 056804 ( 2006 ) . see the supplemental material for details . z. liu , d. l. kovrizhin , and e. j. bergholtz , arxiv:1304.1323 .
discovering pulsars typically involves identifying periodic signals in observational data , then reducing each of them into a set of diagnostic values and graphical representations referred to as a _ candidate_. a modern all - sky pulsar survey such as the high time resolution universe ( htru ; * ? ? ? * ) produces several million such candidates , the overwhelming majority of which are either the result of human - made radio - frequency interference ( rfi ) , or due to various forms of noise . the selection of promising candidates to be observed again for confirmation remains up to this day heavily dependent on human inspection , a very time - consuming process becoming increasingly unmanageable as surveys continue to evolve into ever larger scale operations over time . next generation instruments such as the square kilometre array ( ska ) , can be expected to find 20,000 pulsars @xcite , but not before an estimated 200 million candidates are properly classified , if we are to conservatively assume that the fraction of pulsars to be found among them ( one in ten thousand at most ) is comparable to current surveys @xcite . this implies that , among other challenges , the problem of automated candidate selection must be decisively solved . supervised machine learning ( ml ) classifiers offer great promise in this area , and were first introduced into the field by @xcite . they are general purpose methods that can be used to classify instances of multi - class data , by operating on a well - chosen set of their numerical properties called _ features_. they first build an internal model of the underlying statistical distributions of these features for each data class through the process of _ training_. in the context of supervised learning , this requires a labelled data set carefully prepared by a human expert , or _ training set_. once learned , that internal representation enables the classifier to subsequently label previously unseen data . supervised ml algorithms are well suited to classification problems where no reliable and simple rules are available to perform the task . in this work we use such a class of algorithms , namely artificial neural networks ( ann ) , and attempt to exceed the performance of previous automated candidate selection tools , aiming in particular to correctly label pulsars with a 100% success rate . to achieve that goal , we used a larger training data set that included 1196 pulsar observations from 542 distinct pulsars , wrote a custom ann implementation for increased control over its training process and designed new features to describe the nature of a pulsar candidate . this paper first outlines the problem of candidate selection by visual inspection , and offers a review of existing methods to either reduce the workload or automate the process . in section 3 , we present a detailed introduction to ann . section 4 details the features we use to capture candidate characteristics , and the rationale for their design . in section 5 , our ann implementation is evaluated on a set of candidates from the intermediate galactic latitude area of the htru survey ( htru - medlat ) . its efficiency when used on new data is shown in section 6 . we conclude with a discussion of the reasons of spinn s success , its current limitations , possible future improvements , and how future pulsar surveys should be run with ml classifiers in mind , if human intervention in classification is to be eventually reduced to a bare minimum . identifying new pulsar signals in observational radio data can be done either via single pulse searches @xcite , or periodicity searches which we briefly summarize here . the first computational step , the so - called de - dispersion or dm search , consists in correcting for the dispersive properties of the interstellar medium , which induce a delay in the observed arrival time of pulses that is both dependent on the observational frequency and the a priori unknown free - electron density integrated along the line of sight , a parameter called _ dispersion measure _ ( dm ) . discovering pulsars in binary systems may also require the application of methods to compensate for the effect of orbital motion of the radio source , as its change of velocity along the line of sight causes its apparent pulse period to vary over the course of an observation , as a result of the doppler effect . one of these methods is time domain resampling @xcite , also referred to as acceleration search , working under the assumption that orbital motion during an observation sufficiently shorter than the orbital period of the source is well approximated by a constant acceleration . a thorough processing of the radio data therefore involves a grid search in both dm and acceleration , and for each _ trial _ [ postulated ( dm , acceleration ) pair ] , the time series is transformed accordingly , and periodic signals are identified using a fast fourier transform ( fft ) . finally , the transformed time series can be folded " modulo the period of every significant periodic signal found , coherently stacking and summing the train of pulses of a potential pulsar . the folding process returns the final product of a pulsar survey : candidates with their set of diagnostic information , described shortly after . more detailed information about modern pulsar searching methods can be found in the standard references @xcite . fig.[fig : diagnosticplot ] shows the diagnostic information for a known pulsar that exhibits all the typical characteristics . the plots in the left - hand column describe , from top to bottom , the pulse in different bands of observed frequencies ( sub - bands plot ) , the evolution of the pulse during the observation ( sub - integrations plot ) , and the folded profile " which is the pulse averaged across all the observed frequencies for the entire observation . a pulsar is expected to emit in a broad range of wavelengths , with its signal remaining visible for most of the observation with a stable pulse shape . most pulsars also display a folded profile made of a single narrow peak , although wide and/or multi - peak profiles are not uncommon . the right - hand column of plots contains from top to bottom : the period - dm plane , which represents the evolution of the signal - to - noise ratio ( s / n ) of the signal as it is folded with slightly different values of period and dm . darker colors denote a brighter signal . below , the dm - s / n and acceleration - s / n curves summarize the results of earlier dm and acceleration trials , before the time series was folded , associating the s / n of the candidate in the fourier domain with dm and acceleration trial values . these plots are used to determine that the signal of a prospective pulsar is associated with well - defined and unique values of acceleration , period and dm . for the latter , an unambiguously non - zero value strongly indicates an extra - terrestrial origin for the source . visual inspection of every candidate produced by a modern survey is no longer a reasonable option . as an example , our latest processing of the htru intermediate latitude survey @xcite returned 4.3 million candidates , which , at a very optimistic rate of one candidate per second , would require approximately 1200 person hours to classify . that proposition can be made even less economically interesting in the case of re - processing of data analyzed one or more times before , in which the number of expected new discoveries is much lower . the repetitive nature of the work also leads to errors during long inspection sessions . as a consequence , techniques to reduce the required amount of human intervention have been used for more than a decade . graphical selection tools such as reaper @xcite and jreaper @xcite enable the user to project up to several thousand candidates at once in scatter plots , representing one of their features versus another , leading to rejection en masse of candidates not exhibiting the desired properties , for example excessively faint candidates , or ones found too close to narrow frequency bands polluted by rfi . scoring algorithms such as peace @xcite have also been developed , combining 6 numerical candidate quality factors into one formula that produces a subjective ranking where pulsars are expected to be found close to the top . machine learning ( ml ) solutions have also been proposed , first by @xcite , who used an artificial neural network to classify outputs from the pmps survey , operating on 8 to 12 numerical features extracted from candidate diagnostic information . @xcite applied the same technique on the htru survey , extending the number of features to 22 . more recently , @xcite combined a variety of ml algorithms that perform pattern recognition directly on candidate plots such as those shown in fig . [ fig : diagnosticplot ] , instead of first attempting to reduce them into features . in this work , we use an approach most similar to @xcite . supervised learning is the task of inferring a real - valued function from a set of labelled data points called _ training examples_. a training example consists of a pair @xmath0 where @xmath1 is the input or feature vector , and @xmath2 the _ target value _ or desired output ( typically @xmath3 ) chosen by human experts or gathered from experimental measurements . supervised learning can be used to solve regression problems , where the goal is to predict a continuous variable from a set of inputs , and classification problems , where one tries to assign a discrete class label to new unlabelled data points . in the context of binary classification , the two possible class labels are encoded in the target value , which may be set for example to 1 for members of the positive " class , and to 0 for members of the negative " class . a wide range of supervised learning algorithms is available , including artificial neural networks . one of the valuable features of anns that motivated us to choose them is that they naturally produce a real - valued continuous output , the _ activation value_. while it can be easily converted to a binary class label by applying a threshold , the activation value also represents a level of confidence in the class label obtained . in the context of pulsar candidate classification , this can be used as a way to prioritize inspection and confirmation of candidates as we will see later . in this section we only present an introduction to anns geared more specifically towards their use as binary classifiers . for a more advanced and general overview see e.g. @xcite or @xcite . inputs . the bias term @xmath4 can be seen as a weight operating on an extra constant input equal to one . the activation function @xmath5 usually chosen is the logistic sigmoid or similar.,scaledwidth=45.0% ] an artificial neuron is a computational model ( see fig . [ fig : artificialneuron ] ) inspired by its biological counterpart , which constitutes the basic building block of a network . it is parametrized by a vector of _ weights _ @xmath6 of pre - determined dimension , a scalar _ bias term _ @xmath4 , and an _ activation function _ @xmath5 . for a given feature vector @xmath7 , it outputs an _ activation value _ @xmath8 given by @xmath9 common choices for the activation function are sigmoid shaped non - linear functions such as the hyperbolic tangent or the logistic sigmoid function , the expression of the latter being @xmath10 which takes values between 0 and 1 ( see fig . [ fig : sigmoidfunction ] ) . a very useful geometric interpretation is to visualize a single artificial neuron as defining a separating hyperplane in feature space ( fig . [ fig : neuronhyperplanes ] , top panel ) , with a normal vector @xmath6 defining its orientation , and the bias term @xmath4 defining its altitude at the origin . note that the norm of the weights vector @xmath11 is a meaningful parameter on its own , despite not having any effect on the orientation of the hyperplane : it defines its sharpness " . unsurprisingly , an individual neuron performs poorly on non linearly separable data , but any number of them can be connected into layered networks capable of carving boundaries of arbitrarily high complexity in feature space ( fig . [ fig : neuronhyperplanes ] , bottom panel ) training is the process of finding an adequate set of weights for the given classification problem . this is posed as an optimization problem , where a _ cost function _ or _ loss function _ , which measures the discrepancy between target values and actual outputs of the network on the training set , must be minimized with respect to the weights and biases of the whole network . a common choice of cost function is the mean squared error @xmath12 where @xmath13 and @xmath14 are respectively the weights and bias term of the @xmath15-th neuron in layer number @xmath16 , @xmath17 and @xmath18 are respectively the target value and activation value for training example number @xmath19 , and @xmath20 the total number of training examples . the cost function is minimized using gradient descent , starting from a random initialization and going through iterations where the following three steps are performed in succession 1 . compute the activation values of the network on the data set , and the differences with the target values . 2 . compute the derivative of the cost function with respect to the network parameters ( weights and biases ) , using the _ backpropagation algorithm _ ( see below ) . 3 . correct every network parameter @xmath21 with the following update rule , where @xmath22 is the _ learning rate _ : @xmath23 the backpropagation algorithm ( see e.g. * ? ? ? * for a description and proof ) is a very computationally efficient way to compute the gradient of the cost function with respect to the weights and biases of the network , that historically made the training of large networks tractable . the training process only yields the best network weights and biases to properly label the training set , which does not necessarily imply optimal classification performance on unseen data . the model learned may capture not only legitimate patterns in the data , but also fit irregularities specific to the training set ( due for example to its limited size ) , a situation referred to as _ overfitting_. regularization consists in limiting the complexity of a model to improve its ability to generalize to new data . one such method that we used is _ l2 weight decay _ , where a penalty term is introduced into the neural network cost function usually written as @xmath24 with @xmath25 being the _ weight decay _ parameter . the net effect is to prevent the weights of the network from growing excessively large during training , therefore simplifying the decision boundary shape in feature space . the optimal value of @xmath25 , along with the optimal number of neurons to use , is found through grid search and cross - validation , described in section 5 . features are the properties of an unlabelled data instance upon which a ml algorithm decides to which class it is most likely to belong . the main part of the present work consisted in reducing pulsar candidates into maximally relevant features , i.e. that take values as different as possible for pulsars and non - pulsars , and ensure that these features capture a wide range of information and domain knowledge of a human classifier . to ensure maximum classification performance , particularly with respect to the identification of faint pulsars , we obeyed the following set of guidelines : 1 . reduce selection effects against faint or more exotic pulsars , especially msps or the ones with large duty cycles , which have been the most difficult to identify in the past @xcite . as an example , the number of dm trials or acceleration trials above a certain signal - to - noise threshold were found to introduce a strong and unjustified bias against short period candidates , regardless of their brightness . that feature was used in the past @xcite but not in our own work . 2 . ensure complete robustness to noisy data . as a result , no curve fitting to folded profiles or dm search graphs was attempted , as the results are very difficult to exploit properly in the low s / n regime , in which we are most interested . 3 . limit the number of features to a set of very relevant ones , as the very limited sample of pulsar observations is unlikely to sufficiently cover all the degrees of freedom of a large feature space . an excessive number of features induces a reduction in classification performance . this is a facet of the curse of dimensionality " problem in machine learning known as the _ hughes effect _ @xcite . this also implies avoiding the use of correlated features , as any extra feature must capture additional information . 1 . * signal - to - noise ratio of the folded profile ( log - scale ) . * s / n is a measure of signal significance , which can be defined in various ways . we use the definition ( see e.g. * ? ? ? * ) given by equation . for a given contiguous pulse window @xmath26 , @xmath27 where @xmath28 is the amplitude of the @xmath29-th bin of the folded profile , @xmath30 is the width of the pulse region @xmath26 measured in bins , @xmath31 and @xmath32 are respectively the mean value and the standard deviation of the folded profile in the off - pulse region . the position and width of the pulse are determined by an exhaustive search that maximizes s / n . once determined , the indices of the bins corresponding to pulse and baseline regions are retained in memory for further data processing . we also compute the equivalent width of the profile @xmath33 for further processing , defined by @xmath34 which , in other words , would be the width of a top - hat pulse window that would have the same area and peak height as the original pulse . + since the values of s / n can span a wide range across candidates , we use its logarithm as a feature . past a certain level , an increase of s / n does not make a candidate any more significant to a human expert on its own . all other things equal , two candidates with extremely significant signal - to - noise ratios of 50 and 500 can be considered equally likely to be legitimate pulsars . * intrinsic equivalent duty cycle of the pulse profile . * the duty cycle of a pulsar is the ratio of its pulse width @xmath30 expressed in seconds to its spin period . most folded profiles of pulsars show a narrow pulse with a duty cycle typically below 5% , while a significant amount of terrestrial signals reach much higher values up to 50% . this can often be due to the significant amount of phase drift exhibited by artificial sources during an observation , leading to apparent smearing of their folded profiles . that being said , some pulsars , especially among the millisecond population , can have legitimately large duty cycles , which can be further increased by dispersive smearing . to avoid penalizing such objects , and further increase the usefulness of the duty cycle feature for classification , we remove the effect of dispersive smearing by defining the _ intrinsic equivalent duty cycle _ of a candidate as @xmath35 where @xmath36 is the period of the candidate , @xmath33 its equivalent width defined in expressed in units of time , and @xmath37 the dispersive smearing time across a frequency channel . a first - order approximation of @xmath37 is given by @xmath38 where @xmath39 is the width of an observational frequency channel , @xmath40 the centre observation frequency , and dm the dispersion measure of the candidate . negative intrinsic equivalent duty cycle values are possible , if the dispersive smearing time exceeds the equivalent width . strongly negative values of @xmath41 are not expected for a genuine astronomical signal , and this constitutes an extra selection pattern that can be learned by a ml algorithm . * ratio between barycentric period and dispersion measure ( log - scale ) . * as far as the htru data is concerned , the most pulsar - like rfi candidates ( bright and persistent in time ) tend to appear at periods longer than 1 second , and at dispersion measures close to zero . a way to combine these two selection criteria into one is by considering the ratio between period and dm . as we did for s / n , since the values of this ratio span a large range across the pulsar and rfi population , we actually use the logarithm of the ratio between period and dm as a feature . as shown in fig.[fig : iewvspdm ] , the combination of @xmath42 and intrinsic equivalent duty cycle offers a powerful selection tool that is _ fully independent from s / n _ , splitting clearly the data into three distinct clusters : pulsars , noise candidates ( faint ) , and rfi ( bright ) . note that the usefulness of this feature is dependent on the rfi landscape at the place and even time of observation , and its portability to other surveys is unknown . * validity of optimized dispersion measure . * pulsars can have a wide variety of dispersion measures while their rfi counterparts usually exhibit dm values very close to zero , but no other truly selective pattern based solely on dm can be found . therefore we define the _ validity of dispersion measure _ as @xmath43 the purpose of this feature is to ensure that the classifier learns to very strongly reject candidates with a dm below a certain threshold , below which no pulsars are ever found . we used @xmath44 for htru - medlat data . persistence of signal through the time domain . * a genuine pulsar is expected to be consistently visible during most of an observation , and this provides a selection criterion against impulsive man - made signals . refining an interesting idea proposed by @xcite , we attribute a score " to every sub - integration of the candidate , based on its s / n ( see equation [ eq : snr ] ) measured with respect to the pulse window and baseline region defined by the folded profile . note that negative s / n values are possible if signal is found outside the expected window , a common property of rfi . the scoring function ( see fig . [ fig : scoringfunction ] ) is defined as @xmath45 where @xmath46 is the signal - to - noise ratio of the candidate in a sub - integration , measured as described above , and @xmath4 the benchmark signal - to - noise ratio which is a user - defined parameter . the average of the scores obtained through all sub - integrations constitutes the persistence of the candidate through the time domain . note that @xmath4 should be chosen low enough to filter out signals visible only for a small fraction of the observation , but not so much as to excessively penalize the class of nulling " pulsars @xcite that can become invisible for a part of the observation . we found a sensible , albeit arbitrary choice to be + @xmath47 + with @xmath48 being an estimation of the overall signal - to - noise ratio of the faintest pulsars still clearly visible to the trained eye , and @xmath49 the total number of sub - integrations . we set @xmath50 , also in accordance with the fact that no pulsar discovered with a s / n below 9.5 was ever confirmed over the course of the htru - medlat survey . * root - mean - square distance between the folded profile and the sub - integrations . * the persistence through time " feature is insufficient on its own to capture the information that rfi signals tend to show some amount of drift in phase or even shape changes during an observation , enough to easily betray their non - astronomical nature even to a moderately well - trained human eye . to alleviate this problem , we define a measure of the variability of the pulse shape though the observation . to compute it , we first normalize the folded profile to values between 0 and 1 , and also normalize individually every sub - integration in the same fashion . let @xmath28 be the value of the @xmath29-th bin of the folded pulse profile , and @xmath51 be the value of the @xmath29-th bin of the @xmath15-th sub - integration . let @xmath26 be once more the set of bin indexes that constitute the pulse window in the folded profile , and @xmath30 the pulse window width . we then simply define the root mean square distance between the folded profile of the candidate and each of its sub - integrations as : @xmath52 this feature helps characterize persistent rfi in the medium to high s / n regime . + as a final word , a persistence of signal through the frequency domain " similarly defined as its time domain counterpart was tried and initially believed to provide a very useful selection criterion to separate broadband pulsar signals from all the others . it was eventually removed from the feature set , as its addition proved slightly detrimental to classification performance on htru medlat data . we attributed this effect to the relative absence of candidates that arise from narrowband radio frequency interference , but it might be useful for other surveys facing different rfi populations . computing persistence features as defined earlier relies on comparing the brightness of the signal vs. a benchmark , which is constant in the time domain . for surveys with large fractional bandwiths , correcting for the average spectral index of pulsars when computing that benchmark in the frequency domain could be beneficial . training examples were gathered from the outputs of a new processing of htru medlat data , performed as a test run of the new high - performance gpu - based peasoup pipeline @xcite , with acceleration searching enabled . dms from 0 to 400 @xmath53 , and accelerations from @xmath54 to @xmath55 @xmath56 were searched , yielding in excess of a thousand candidates in each of the 95,725 beams . signal periods simultaneously found in three beams or more of the same pointing were ignored , and the resulting 50 brightest candidates in every beam were folded , with the exception that any signal with a fourier - domain s / n in excess of 9 was automatically folded as well . this processing strategy returned 4.34 million folded candidates . each of them was individually matched against the atnf pulsar catalogue @xcite to label all known pulsars ( and their harmonics ) that were found by the pipeline . a total of 1196 observations of 542 distinct known pulsars were identified after being carefully reviewed by eye to confirm their nature . they constituted the positive class of the ann training set , to which we added 90,000 non - pulsar observations picked at random as a negative class , to obtain a varied and representative sample of the population of spurious folded candidates . we assumed that none of these were undiscovered pulsars : in past htru medlat processings , approximately 100 new pulsars were found in 10 million folded candidates , a discovery rate of 1:100,000 . in our re - processing set of candidates , it is expected to be significantly lower . the training set contains pulsars with varied spin periods , duty cycles , and signal - to - noise ratios . among them , 77 have periods shorter than 50 ms and 46 have duty cycles larger than 20% . a total of 78 observations of pulsars have folded signal - to - noise ratios below 10 , down to a minimum of 7.3 . for the sake of comparison , no pulsar discovered with a folded s / n below 9.5 was ever confirmed over the entire course of the htru survey . low s / n pulsars are difficult to distinguish from noise fluctuations even by eye , and the limited observation time available imposes a conservative s / n selection threshold ( a consensual value is 10 ) on the folded candidates to confirm . despite these limitations , fainter pulsar observations were kept in the training set to ensure maximum sensitivity . to obtain fine control over the training process , a custom ann implementation was written . we followed some practical recommendations detailed at length in @xcite , which we enumerate here . 1 . feature scaling was performed before training , ensuring that every individual feature has zero mean and unit standard deviation over the entire training set . the hyperbolic tangent activation function was used instead of the logistic sigmoid , yielding activation values ranging from @xmath57 to @xmath58 . the ann was trained using mini - batches " , whereby during each training epoch the network is presented only with a small , changing subset of the training set . this is not only a much faster process than standard batch " training , but also its noisy nature can help to avoid local minima of the network cost function . to overcome the large class imbalance of the training set , we oversampled the pulsars to obtain a 4:1 ratio of non - pulsars to pulsars , so that pulsar candidates were seen " much more often by the ann during training , while preserving the variety of non - pulsars . obviously , the goal of an automated classifier is to identify the largest possible fraction of pulsars while returning a minimal amount of mislabelled noise or rfi ( false positives ) . the two natural performance metrics for this problem , measured on a test sample of labelled data , are therefore : @xmath59 @xmath60 the true positives are the positive examples correctly labelled as such , and false negatives are the positive examples incorrectly labelled as negatives . recall is therefore the fraction of positives properly labelled . likewise , the false positive rate is the fraction of negatives mislabelled as positives . other common metrics such as _ accuracy _ or _ f - score _ depend explictly on class imbalance ( the ratio of positive to negative examples in the test data ) , and as such are not suitable if we are to compare classifier performances across different test samples or even different pulsar surveys . since pulsars remain arguably rare objects up to this day , missing any of them carries a heavy cost , and emphasis must be put on maximising recall above all else . the amount of visual inspection required to select folded candidates for confirmation is proportional to the false positive rate of the classifier , for which low values are desirable . we performed a 5-fold cross - validation procedure , to choose an optimal network architecture and weight decay , and evaluate classification performance . this consists in randomly partitioning all of the labelled data into 5 equally sized subsets , each of them being successively held out as a test set , while only the remaining four are used to train the ann upon . this ensures that performance is always evaluated on data unseen during training , and that over the five iterations ( folds " ) of this procedure every candidate ends up in the test set exactly once , at which point its ann activation value or score " ( that takes continuous values between -1 and + 1 ) was recorded . we repeated the entire process 20 times , with different random data partitions , obtaining a representative average score for every candidate from our labelled sample that does not depend on a specific training / test set split . the resulting list of scores allowed to determine , for every score decision threshold between @xmath57 and @xmath58 , what were the associated recalls and false positive rates , ie . how many pulsars were below that decision threshold , and how many rfi or noise candidates were above . when later deploying the ann on new data , this gives an estimate of how far down the score ladder candidates should be inspected by eye , where the tradeoff between recall and false positive rate is left to the user s discretion . furthermore , examining consistently low scoring pulsars gave insight into which ones the ann was biased against , which we will discuss later . finally , by repeating this whole procedure with various network architectures and weight decay values , we were able to settle on an optimal ann configuration . it was chosen so as to minimize the number of non - pulsars scoring better than the worst - scoring pulsar , that is minimize the false positive rate at 100% recall . a simple two - layered 8:1 network ( 8 hidden units , one output unit ) was found to yield the best results , with performance progressively degrading with more units . [ fig : xvresults ] shows the distribution of scores obtained by all candidates in our labelled sample during cross - validation , and illustrates the recall / false positive rate tradeoff . table [ table : performancesummary ] summarizes expected classification performance for various score decision thresholds . the score distribution of pulsars shows a long tail where no more than a dozen low - scoring pulsars are responsible for the major part of the false positive rate . their close inspection reveals that they always share at least two of the following characteristics , making them similar to noise or rfi candidates with respect to the feature space we used : large duty cycles in excess of 20% , low s / n ( below 9 ) , high value of @xmath61 . a few other low - scoring pulsar observations were found to exhibit abnormally low persistence through time , being rendered invisible during a part of the observation by short bursts of rfi . .classification performance during cross - validation , summarizing some key values from fig . [ fig : xvresults ] . [ cols="<,>,>",options="header " , ] [ table : discoveries ] a fully trained ann was deployed on all 4.34 million candidates returned by the processing previously described in section 5.1 , a process that takes only 400 cpu - hours , despite being severely limited in speed by a large amount of small file i / o operations . using 64 cpus on swinburne university s gstar cluster , this can be done overnight . candidates were sorted by decreasing score , and known pulsars and their harmonics were removed from the list . in light of the cross - validation results ( see fig . [ fig : xvresults ] and table [ table : performancesummary ] ) , one can reasonably expect to find all potential discoveries above a score threshold of @xmath62 , which left approximately 27000 candidates to review . [ subsec : visualinspectionreport ] so far all 2400 candidates that scored above @xmath63 returned by spinn have been inspected , a process that will continue as observation time to confirm possible discoveries becomes available . table [ table : discoveries ] summarizes the attributes of the most promising candidates found among them . it shows that spinn is very sensitive to pulsar - like signals down to s / n = 8 , and that it can also highly rank broad pulses ( duty cycles in excess of 20% ) and potential millisecond pulsars , which is a known blind spot of some previous ml solutions @xcite . it should be noted that a considerable amount of the reviewed candidates were rfi signals with very specific periods . [ fig : logperioddistribution ] presents the distribution in @xmath64 of all previously unknown 7094 candidates that obtained a positive score . one percent of the bins account for 50% of these candidates . one could either postpone or skip the inspection of heavily polluted period intervals , or adjust _ a posteriori _ all ann scores via bayesian inference , using the period distribution of high scoring candidates and that of known pulsars @xcite . we intend to implement such a scheme in the future . the candidates of table [ table : discoveries ] with a s / n above 9.5 have been reobserved at the parkes observatory and four were confirmed as new pulsars , three of which have spin periods shorter than 5 ms . [ fig : discoveries ] shows their candidate plots exactly as they were evaluated by spinn . the details of these four new discoveries will be discussed in a future paper @xcite once their long - term coherent timing solutions have been obtained from currently ongoing observations . we have described spinn , an automated pulsar candidate classifier designed with maximum recall in mind . being essentially an artificial neural network that produces a real - valued and continuous output instead of a binary class label , it can also produce a subjective ranking of candidates that can be used to prioritize visual inspection . spinn was cross - validated on a data set containing all known pulsars found by the peasoup pipeline in a re - processing of htru - medlat and 90,000 non - pulsar candidates chosen at random . its expected recall and false positive rates were evaluated ( see table [ table : performancesummary ] ) and it was found to be capable of reducing the survey s outputs by a factor of approximately 150 while identifying all potential pulsars . reduction factors of several thousand can be achieved at the cost of postponing a small fraction of new discoveries , an interesting prospect for future surveys . spinn was deployed on all candidates produced during re - processing and four new pulsars were discovered in the 2,400 candidates it ranked most highly ( less than 0.06% of the survey s output ) . three of them are millisecond pulsars , one of which was found with a signal - to - noise ratio below 10 . while spinn seems to be a significant step forward compared to previous solutions , the performance of automated candidate classifiers depends very significantly on the properties of the data they are evaluated upon . the amount of bright spurious candidates to be sifted through will be affected by the rfi landscape at the observation site , and rfi mitigation techniques used during early stages of observational data processing . the available quantity and variety of known pulsar candidates to train ml algorithms upon also plays a role , and the absence of even a handful of pulsars difficult to detect even by visual inspection in test data can skew the results heavily . [ fig : xvresults ] illustrates this fact , as the removal of ten well - chosen " pulsars from our training data could have unjustifiably reduced the reported false positive rate by an order of magnitude , while obviously reducing spinn s sensitivity . comparisons between automated solutions are therefore limited at best , unless they are made on a common data set . to address this issue , we have made publicly available the set of candidates on which spinn was cross - validated ( see appendix [ sec : trainingdata ] ) . this will allow other authors of classifiers to evaluate their own solutions , and hopefully stimulate interest in the pulsar candidate classification problem , even from machine learning enthusiasts not necessarily acquainted with astronomy . a summary of reported performance of existing automated candidate classifiers is provided in table [ table : performancecomparison ] with all the previous caveats in mind . lllllr classifier & type & recall & false positive rate & test data origin & comments + @xcite & ann & 85% & 1% & htru@xmath65 medlat & + & & 93% & 1% & & 12:2 network + & & 92% & 0.5% & & 8:2 network + & & 100% & 3.7% & & + & & 95% & 0.34% & & + & & 68% & 0.17% & & + & & 92% & 1% & palfa@xmath66 & + & & 100% & 3.8% & gbncc@xmath67 & trained on palfa data + & & 100%@xmath68 & 1.1%@xmath68 & gbncc@xmath67 & trained on palfa data + & & 68% & 0.16% & gbncc@xmath67 & trained on palfa data + & & 100% & 0.64% & & + & & 99% & 0.11% & & + & & 95% & 0.01% & & + + + + + + [ table : performancecomparison ] ml algorithms operate on numerical features that carry no label or context , and unlike human classifiers , can not rely on any domain knowledge associated with these features . therefore , the statistical distributions of these features for different classes ( pulsars and non - pulsars ) should overlap as little as possible , so that these classes are more easily separated in feature space ( see fig . [ fig : neuronhyperplanes ] for a visual interpretation on a toy example ) . different features were tried and discarded in the context of this work and the best subset of them , reported in section [ sec : features ] , selected through cross validation with maximum recall in mind . also , while this may appear counter - intuitive , ml algorithms do not always perform better with more features . for a given amount of training data , larger feature spaces will be more sparsely sampled , and the data distributions more difficult to infer accurately . the decision boundaries separating these distributions must be of appropriate complexity , which , in the case of artificial neural networks , increases with the number of layers and neurons in the network . we attribute part of spinn s success to its relatively simple internal model of a pulsar candidate . the low number of features and neurons used are in accordance with the limited amount of pulsar observations available . this can be quantitatively supported by the fact that the pics ( without score adjustment ) and peace classifiers show almost identical performance on candidates taken from gbncc data , as shown in table [ table : performancecomparison ] . pics makes use of a committee of ml classifiers , including deep neural networks containing in excess of eight thousand neurons , while peace relies on a linear function of six numerical features . in this case , the large increase in model complexity did not translate into a significant improvement of classification performance . spinn s limitations are closely related to the features it relies upon . the @xmath69 feature has provided a very simple and efficient selection criterion against artificial signals on htru - medlat data partly because a large majority of rfi appears at periods in excess of one second ( fig . [ fig : logperioddistribution ] ) , a rule not guaranteed to hold true in other surveys . cross - validation also indicated that spinn carries a bias against pulsar signals showing a large duty cycle , and scores them even lower if they also exhibit a high @xmath69 value , as any pulsar signal with these two properties becomes difficult to distinguish from spurious ones in spinn s feature space ( fig . [ fig : iewvspdm ] ) . characterization of rfi is therefore largely incomplete , which was confirmed by the visual inspection of high scoring candidates ( see section [ subsec : visualinspectionreport ] and table [ table : discoveries ] ) . with the known pulsars and their harmonics ignored , the remaining shortlist was dominated by artificial signals that had pulsar - like properties with respect to spinn s feature space , yet their nature was often recognizable to the eye . some signals were found to drift erratically over time , or to emit only in a set of narrow frequency bands . while spinn has been successful in finding new pulsars , its false positive rate would have to be further reduced by about two orders of magnitude for it to be a match for the human eye in terms of accuracy , and while this is an ambitious goal , it leaves the problem of automated pulsar candidate classification largely open . we have previously suggested that spinn s success comes from its low - complexity model of a pulsar candidate , relying on a lower number of features and neurons than what has been previously used , a consequence of the limited number of known pulsars observable from any given site on earth . this would have two main implications for any present or future ml solutions . firstly , collecting as many pulsar observations as possible for training should be a top priority . observing known sources multiple times could be a viable option , purposely not pointing exactly at them to simulate a blind all - sky search , or purposely processing only a part of such an observation to reduce the s / n of the output candidate . artificially generating credible pulsar candidates is another possibility previously proposed several times @xcite . this would come with the challenges associated with realistically simulating the varying properties of pulsar signals and all forms of noise or interference that affect them . secondly , additional efforts should be undertaken to improve the quality of the folded data if we are to increasingly rely on ml for pulsar candidate selection . dealing with candidates folded with wrong periods , dispersion measures or accelerations vastly increases the number of degrees of freedom of the classification problem by perturbating the candidate plots in various ways . while such mistakes can be easily spotted by eye with domain knowledge and the wealth of information available in candidate plots , ml algorithms would require more features to achieve the same , and enough training data to properly sample the space of possible processing errors . instead , extra features that can be afforded would find better use in improving rfi characterization , which will be the main focus of our future efforts to further increase classification performance . the parkes observatory is part of the australia telescope which is funded by the commonwealth of australia for operation as a national facility managed by csiro . this work was supported by the australian research council centre for excellence for all - sky astrophysics ( caastro ) , through project number ce110001020 . the re - processing and scoring of the htru medlat survey made extensive use of the gpu supercomputer for theoretical astrophysics research ( gstar ) hosted at swinburne university , and funded by a grant obtained via astronomy australia limited ( aal ) . we thank the anonymous referee for their helpful remarks . the entire training data set that was used to cross - validate spinn in this work is available at http://astronomy.swin.edu.au/~vmorello/	we describe spinn ( straightforward pulsar identification using neural networks ) , a high - performance machine learning solution developed to process increasingly large data outputs from pulsar surveys . spinn has been cross - validated on candidates from the southern high time resolution universe ( htru ) survey and shown to identify every known pulsar found in the survey data while maintaining a false positive rate of 0.64% . furthermore , it ranks 99% of pulsars among the top 0.11% of candidates , and 95% among the top 0.01% . in conjunction with the peasoup pipeline @xcite , it has already discovered four new pulsars in a re - processing of the intermediate galactic latitude area of htru , three of which have spin periods shorter than 5 milliseconds . spinn s ability to reduce the amount of candidates to visually inspect by up to four orders of magnitude makes it a very promising tool for future large - scale pulsar surveys . in an effort to provide a common testing ground for pulsar candidate selection tools and stimulate interest in their development , we also make publicly available the set of candidates on which spinn was cross - validated .
the nature of the interaction between the hydrogen bond ( hb ) network of liquid water and the solvated chloride ion ( a member of the hofmeister series ) is currently a topic under intense research due to its fundamental importance in biochemistry , atmospheric chemistry , and geological processes @xcite . experimentally , the interaction between the solvated chloride ion , cl@xmath1 , and the surrounding aqueous environment has been successfully probed by a variety of techniques @xcite . in particular , x - ray and neutron scattering as well as x - ray absorption measurements , have been used to investigate the solvation structure of aqueous chloride solutions , providing experimental determination of the cl - o and cl - h radial distribution functions ( rdf ) , @xmath2 and @xmath3 , and the associated coordination number , _ i.e. _ , the number of water molecules populating the first solvation shell surrounding a given cl@xmath1 ion @xcite . in addition , the electronic structure associated with this fundamental aqueous ionic solution was studied by delahay @xcite and winter _ et al . _ @xcite , in which state - of - the - art photoemission spectroscopy ( pes ) was utilized to characterize the valence 3@xmath4 bands of the solvated cl@xmath1 ion , which were found to be approximately 1.251.50 ev above the valence band maximum ( vbm ) of liquid water . from a theoretical and computational point of view , first - principles based computational methods such as _ ab initio _ molecular dynamics ( aimd ) have become powerful tools in the study of condensed - phase systems such as liquids @xcite and aqueous ionic solutions @xcite . with the aimd technique , the nuclear potential energy surface is generated `` on the fly '' from the electronic ground state @xcite without the need for empirical input , thereby allowing for a quantum mechanical treatment of not only the structure and dynamics of a given molecular system of interest , but also its electronic and dielectric properties , as well as potential chemical reactions @xcite ( _ i.e. _ , the breaking and forming of chemical bonds ) . since the initial pioneering simulations of liquid water , @xcite aimd has been applied to many complex problems in biology , chemistry , and energy research , _ e.g. _ , the designing of efficient catalysts for hydrogen production @xcite and the modeling of the auto - ionization @xcite and electrocatalytic splitting of water , @xcite to name a few . in particular , aimd simulations employing density functional theory ( dft ) as the source of the underlying quantum mechanical potential have been used to generate reasonably accurate microscopic descriptions of the structure of the aqueous chloride ion solution , @xcite with rdfs and associated coordination numbers that were consistent with the available experimental data . these studies found that the predicted solvation structure in the aqueous chloride ion solution is moderately dependent on the exchange - correlation ( xc ) potential , with semi - local generalized - gradient approximation @xcite ( gga ) functionals such as pbe @xcite and blyp @xcite yielding similar predictions to hybrid functionals such as pbe0 . @xcite in addition , the hb network of water was found to be only locally perturbed by the presence of the solvated ion , a finding which is consistent with the general expectation that cl@xmath1 is a weak disrupter of the aqueous environment @xcite . however , these studies also demonstrated that the electronic structure of this aqueous ionic solution is very sensitive to the adopted xc approximation @xcite . in particular , higher - level electronic structure calculations performed on gga - dft generated configurations ( structures ) were found to significantly underestimate the energy levels of the solvated cl@xmath1 ion , with resulting energetics that can even be qualitatively incorrect @xcite . in the study of an aqueous ionic solution , an accurate theoretical description of the underlying hb network is a prerequisite for characterizing and understanding the interaction between a solvated or embedded ion and its surrounding environment . in neat liquid water , the predictive power of dft - based aimd in the microscopic description of the hb network critically depends on the accuracy of the underlying xc functional utilized in the quantum mechanical treatment of the electronic degrees of freedom . in this regard , it is now clear that the most widely used xc functionals , _ i.e. _ , the class of functionals based on the gga , have severe limitations when applied to liquid water @xcite as well as the crystalline phases of ice . @xcite most notably , gga - dft suffers from the presence of self - interaction error ( sie ) @xcite and the neglect of non - local electron correlation effects that are responsible for van der waals ( vdw ) or dispersion interactions ; as a result , these deficiencies in the gga xc potential manifest as significant overstructuring and excessively sluggish dynamics in aqueous systems such as ambient liquid water . beyond the choice of the xc functional , most aimd simulations of liquid water and aqueous ionic solutions performed to date have adopted classical mechanics for the nuclear equations of motion and therefore completely neglect nuclear quantum effects ( nqe)another approximation that has been deemed insufficient for a highly accurate quantitative description of the microscopic structure and hb network in aqueous systems . in the case of liquid water , light atoms such as hydrogen deviate significantly from classical behavior even at room temperature , @xcite as evidenced by experimental isotope effect studies which demonstrated a softening of the liquid structure ( _ i.e. _ , in the comparison of h@xmath5o to d@xmath5o ) . @xcite hence the neglect of nqe in aqueous systems such as ambient liquid water leads to overstructuring in the predicted rdfs . @xcite a commonly adopted method to alleviate the deleterious effects of sie in gga - dft is the use of hybrid xc functionals , wherein a fraction of exact ( or hartree - fock ) exchange ( @xmath0 ) is included in the density functional approximation . due to the relatively high computational cost associated with these xc functionals , applications of hybrid dft have mostly been restricted to small gas - phase clusters of water , @xcite although recently hybrid functionals have been applied in the study of several crystalline phases of ice @xcite and liquid water . @xcite in comparison to ggas , these studies demonstrated that the energetic , structural , and vibrational properties of these aqueous systems , as predicted by hybrid dft calculations , are generally in closer agreement with the available experimental data . @xcite indeed , an accurate microscopic description of the hb network by hybrid xc functionals has a non - negligible effect on the theoretical characterization of the solvated chloride ion . in this regard , it was found that the use of hybrid dft was crucial in obtaining a qualitatively correct energy difference between cl@xmath1 and the vbm of liquid water . in addition , both hybrid and gga xc functionals lack the ability to describe vdw / dispersion interactions , which arise from non - local dynamical electron correlation and have a substantial effect on the microscopic structure of condensed - phase aqueous systems . in fact , the explicit inclusion of vdw interactions in dft has been shown to significantly improve upon the theoretical description of the transition pressures among the high - pressure phases of ice @xcite and the predicted equilibrium density of liquid water . @xcite while many recent studies have concluded that the structure of liquid water significantly softens when vdw interactions are accounted for in the underlying xc potential , the extent to which these non - local forces affect the structure of liquid water is largely dependent upon the given approach utilized to facilitate vdw - inclusive dft . @xcite hence , the inclusion of vdw or dispersion interactions in the underlying xc potential is expected to have a non - negligible effect on the theoretical description of the hb network and must be accounted for in the study of aqueous ionic solutions such as the solvated chloride ion . recently it was shown that utilization of the hybrid pbe0 functional , @xcite which includes 25% exact exchange , in conjunction with a fully self - consistent ( sc ) implementation of the density - dependent vdw / dispersion correction of tkatchenko and scheffler @xcite ( ts - vdw ) , _ i.e. _ , the pbe0+ts - vdw(sc ) xc functional , yields an oxygen - oxygen structure factor , @xmath6 , and corresponding rdf , @xmath7 , that are in quantitative agreement with the best available experimental data . @xcite this level of agreement between _ ab initio _ simulations and experiment was attributed to an increase in the relative population of water molecules in the interstitial region ( _ i.e. _ , the region between the first and second coordination shells ) , a collective reorganization in the liquid phase which is facilitated by a weakening of the hb strength by the use of a hybrid xc functional , coupled with a relative stabilization of the resultant disordered liquid water configurations by the inclusion of long - range vdw / dispersion interactions . in fact , this increasingly more accurate description of the underlying hb network in liquid water also yielded other correlation functions , such as the oxygen - hydrogen rdf , @xmath8 , and the higher - order oxygen - oxygen - oxygen triplet angular distribution , @xmath9 , which encodes the degree of local tetrahedrality , as well as electrostatic properties , such as the effective molecular dipole moment , that are in much better agreement with experiment . @xcite in this regard , the overall agreement between experiment and the pbe0+ts - vdw(sc ) description of the microscopic structure of ambient liquid water is indeed a very promising starting point for accurately investigating the structural and energetic properties that characterize the aqueous chloride ion solution , which is the main focus of the work reported herein . in this work , we have systematically performed a series of car - parrinello aimd simulations @xcite of the aqueous chloride ion solution at ambient conditions using a hierarchy of different xc functionals . the sequence of xc functionals employed herein includes the standard semi - local gga of perdew , burke , and ernzerhof ( pbe ) , @xcite the corresponding hybrid pbe0 @xcite which includes 25% exact exchange , and the self - consistent ( sc ) dispersion - corrected analogs @xcite thereof , _ i.e. _ , pbe+ts - vdw(sc ) and pbe0+ts - vdw(sc ) , based on the tkatchenko - scheffler @xcite density - dependent vdw / dispersion functional . all of these aimd simulations were performed in the canonical ( @xmath10 ) ensemble using periodic simple cubic simulation cells containing cl@xmath1 ion in ( h@xmath5o)@xmath11 with lattice parameters set to reproduce the experimental density of liquid water at ambient conditions . all of the aimd simulations were initially equilibrated for approximately 6 ps and then continued for an additional 2050 ps for data collection ( with the aimd simulations employing the hybrid pbe0 functional at the lower end of this range of simulation times ) . since a classical treatment of the nuclear degrees of freedom was found to be insufficient for a quantitatively accurate description of the microscopic structure of ambient liquid water , we have performed all aimd simulations in this work at the elevated temperature of 330 k , a technique that is suggested by the lowest - order perturbative expansion of the free energy ( in @xmath12 ) to account for the quantum mechanical nature of the nuclear degrees of freedom . @xcite in practice , this increase of approximately 30 k in the simulation temperature has been found to mimic the nuclear quantum effects ( nqe ) in structural quantities such as the oxygen - oxygen radial distribution function ( @xmath7 ) in both dft @xcite and force field @xcite based md simulations of liquid water . all calculations reported herein were performed within the plane - wave and pseudopotential framework and utilized a modified development version of the quantum espresso ( qe ) software package . @xcite to meet the additional computational demands associated with large - scale aimd simulations based on hybrid xc functionals , we have employed a linear scaling o(@xmath13 ) exact exchange algorithm that exploits the natural sparsity associated with the real - space maximally localized wannier function ( mlwf ) @xcite representation of the occupied kohn - sham electronic states . @xcite in addition , we have also developed and utilized a linear scaling o(@xmath13 ) self - consistent implementation of the ts - vdw dispersion correction , @xcite which provides a framework for computing atomic @xmath14 dispersion coefficients as explicit functionals of the charge density , _ i.e. _ , @xmath15 $ ] , thereby accounting for the local chemical environment surrounding each atom . @xcite more explicit descriptions of the theoretical methods employed herein can be found in ref . [ ] . the car - parrinello ( cp ) @xcite equations of motion for the nuclear and electronic degrees of freedom were integrated using the standard verlet algorithm and a time step of 4.0 a.u . ( @xmath16 0.1 fs ) . to ensure an adiabatic separation between the electronic and nuclear degrees of freedom in the cp dynamics , we used a fictitious electronic mass of 300 a.u . , which was found to be a reasonable choice for the simulation of water , @xcite and the nuclear mass of deuterium for each hydrogen atom . all electronic wavefunctions were expanded using a plane wave basis set with a kinetic energy cutoff of 72 ry , and the interactions between the valence electrons and the ions ( consisting of the nuclei and their corresponding frozen - core electrons ) were treated with troullier - martins type norm - conserving pseudopotentials . @xcite ionic temperatures were controlled with massive nos - hoover chain thermostats , @xcite each with a chain length of 4 . @xcite a neutralizing background charge was included in the ewald summation of the electrostatic energy to compensate for the net negative charge of the chloride ion solution . to examine the electronic structure surrounding the solvated chloride ion , we have performed wavefunction optimizations ( _ i.e. _ , self - consistent solutions of the non - linear kohn - sham equations ) using the pbe , pbe0 ( 25% exact exchange ) , and bhlyp @xcite ( 50% exact exchange ) xc functionals on a set of configurations ( structures ) selected at even intervals along a given md trajectory . real - space integration of the orbital densities surrounding the solvated chloride ion was performed within a sphere of radius @xmath17 to determine the ionic contribution to the overall electronic structure and to identify the highest - occupied molecular orbital ( homo ) states . all pbe and pbe0 wavefunction optimizations were performed with the qe software package . @xcite calculations with the bhlyp xc functional were performed with the cp2k suite of programs @xcite using a split - valence triple-@xmath18 basis set ( appended with an additional set of polarization functions ) in conjunction with goedecker - teter - hutter type pseudopotentials @xcite and a density cut - off of 400ry . and @xmath2 , of the aqueous chloride ion solution obtained from theory _ via _ the dft - based aimd simulations performed in this work.,width=302 ] [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] we now focus our attention on the electronic properties of the aqueous cl@xmath1 ion solution . in particular , @xmath19 , which is defined as the relative energetic difference between the solvated cl@xmath1 ion and the vbm of liquid water will play a key role in our analysis . in this regard , it was previously found that the energy level of the solvated cl@xmath1 ion is quite sensitive to both the underlying molecular structure as well as the choice of the electronic structure method employed during ground state wavefunction optimization . @xcite therefore , a systematic study that differentiates these two effects is necessary and will be addressed in this work . ion solution computed using the hybrid pbe0 xc functional based on structures ( configurations ) generated from aimd trajectories at the ( a ) pbe , ( b ) pbe+ts - vdw(sc ) , ( c ) pbe0 , and ( d ) pbe0+ts - vdw(sc ) levels of theory . the dos and pdos above were obtained from electronic structure calculations on 200 configurations taken from each aimd trajectory.,width=302 ] . [ fig : all - pbe0 ] to accomplish this goal , we first performed band structure calculations with the hybrid pbe0 xc functional based on structures ( configurations ) generated with aimd simulations at the pbe , pbe+ts - vdw(sc ) , pbe0 , and pbe0+ts - vdw(sc ) levels of theory . with a self - consistent solution to the non - linear kohn - sham ( ks ) equations , we projected the optimized ground - state ks eigenfunctions onto the solvated cl@xmath1 ion with a radius of @xmath17 . @xcite the resulting total density of states ( dos ) and real - space projected dos ( pdos ) computed at the pbe0 level of theory are plotted in fig . [ fig : all - pbe0 ] , from which it is evident that the total dos of the aqueous cl@xmath1 ion solution is dominated by the features belonging to the valence electrons of liquid water . in this regard , the oxygen @xmath20 states are located well above ( @xmath21 ev ) the oxygen semi - core @xmath22 states , and can be further decomposed into three features , corresponding to the @xmath23 , @xmath24 , and @xmath25 bands representing the covalently bonded and non - bonded ( lone pair ) electrons . as far as the pdos of the solvated cl@xmath1 ion is concerned , the @xmath26 states of the cl@xmath1 ion were found to be between the @xmath22 and @xmath20 bands of liquid water while the @xmath27 states of the cl@xmath1 ion were found to be located at the edge of the vbm of liquid water a picture which is consistent with the pes measurements in aqueous cl@xmath1 ionic solutions . @xcite moreover , it can be seen that the total dos computed based on molecular configurations generated from different aimd trajectories remain very similar in terms of the peak positions and spectral distributions . this finding is indicative of the fact that the band structure of liquid water is rather insensitive to the underlying molecular configurations ( _ i.e. _ , to whether or not @xmath0 and/or non - local vdw / dispersion interactions are accounted for in the underlying xc potential ) . in sharp contrast , the relative position of the orbital levels associated with the solvated cl@xmath1 ion with respect to the vbm of liquid water shows a much stronger dependence on the underlying molecular configurations . as plotted in fig . [ fig : all - pbe0 ] , the pdos computed at the pbe0 level of theory based on molecular configurations generated from the pbe - aimd trajectory is mostly located below the vbm of liquid water , yielding a value of @xmath28 ev ( see table [ table2 ] ) . with exact exchange and/or non - local vdw / dispersion interactions accounted for in the underlying potential to generate the aimd trajectories , the distribution of the pdos becomes more prominantly separated from the vbm of liquid water . as a result , the computed @xmath19 increases in the direction of the experimental value of @xmath29 1.251.50 ev as shown in table [ table2 ] , a result that we attribute to the weakened interaction between the cl@xmath1 ion and its surrounding water molecules with the use of vdw - inclusive hybrid xc functionals . as discussed above in sec . [ sec : solvation ] , the collective effects of @xmath0 and non - local vdw / dispersion interactions lead to a net decrease in the strength of the hydrogen bonds existing between the lone pair electrons of the cl@xmath1 ion and the water molecules residing in the first solvation shell . with the positions of the first maximum in @xmath2 and @xmath3 , _ i.e. _ , @xmath30 and @xmath31 , slightly increased , hybridization between the cl@xmath1 @xmath27 and the oxygen @xmath20 orbitals ( located at the edge of the liquid water valence band ) becomes less favorable . we note in passing that this explanation is consistent with the increased @xmath19 depicted in fig . [ fig : all - pbe0 ] . ion solution computed using the ( a ) pbe , ( b ) pbe0 , and ( c ) bhlyp xc functionals based on structures ( configurations ) generated from an aimd trajectory at the pbe0+ts - vdw(sc ) level of theory . the dos and pdos above were obtained from electronic structure calculations on 200 configurations taken from the aimd trajectory.,width=302 ] from the above discussion , one can see that an accurate prediction of the electronic properties of the solvated cl@xmath1 ion crucially depends on the description of the interaction between the @xmath4 electrons of cl and o. in this regard , the more accurate prediction of the hydrogen bond tends to discourage hybridization among these @xmath4 orbitals , thereby increasing @xmath19 towards the experimental value . this trend is even more pronounced when different xc approximations are utilized in the theoretical treatment of the electronic structure . as shown in fig . [ fig : pbe0-vdw ] , we now present the dos and pdos of the aqueous cl@xmath1 ion solution based on ground - state wavefunction optimizations performed at the pbe , pbe0 , and bhlyp levels of theory . in this case , the underlying molecular structures or configurations are kept fixed and were taken from an aimd trajectory employing the vdw - inclusive hybrid pbe0+ts - vdw(sc ) xc potential . this choice is motivated by our recent study in which we found that the pbe0+ts - vdw(sc ) xc potential yielded a very accurate prediction of the microscopic structure of ambient liquid water that is in quantitative agreement with respect to the available scattering experiment data . @xcite as clearly seen from fig . [ fig : pbe0-vdw ] , the pdos of the solvated cl@xmath1 ion is significantly affected by the choice of dft functional employed in the electronic structure calculation . at the semi - local pbe - gga level of theory , the computed value of @xmath32 ev is negative , which is opposite in sign compared to the experimental value ; however , as the fraction of @xmath0 is increased from 25% ( pbe0 ) to 50% ( bhlyp ) , the @xmath19 steadily increases to 0.34 ev and 0.92 ev , respectively , and is rapidly approaching the experimental value . again , we attribute this trend to the removal of the deleterious self - interaction error that is present in semi - local xc functionals _ via _ the admixture of exact exchange in the hybrid pbe0 and bhlyp xc functionals . as such , the electronic structure predicted by semi - local gga functionals ( such as pbe ) artificially favors the interaction of the cl@xmath1 ion and its surrounding water molecules by facilitating a greater extent of @xmath4 orbital hybridization . this effect is greatly reduced at the hybrid dft level of theory at the bhlyp level , the eigenfunctions of the cl@xmath1 ion homo were found to have a non - negligible amplitude located on water molecules in the first coordination shell in only @xmath16 5% of the configurations . however , we note in passing that this fraction increases to over 70% when the semi - local pbe - gga xc functional is used instead ( see fig . [ fig : orbital ] ) . orbitals for the cl@xmath1 ion.,width=302 ] these findings reported above are consistent with the previous study by zhang _ et al . _ @xcite in which it was found that the use of hybrid xc functionals in generating both the electronic and molecular structure was essential for a qualitatively correct prediction of @xmath19 . the current study further indicates that accounting for non - local vdw / dispersion interactions in the generation of the underlying molecular structure is also important for obtaining a quantitatively correct theoretical prediction of this fundamental quantity in aqueous ionic solutions . by admixing 25% of @xmath0 , the hybrid pbe0 xc functional has successfully generated an electronic structure that is qualitatively consistent with the available pes experiments . however , 25% of @xmath0 might not be sufficient to completely eliminate the effects of self - interaction error on @xmath19 , @xcite which can serve as an explanation for the remaining discrepancy that exists between theory and experiment . in addition , it should also be noted that an explicit treatment of nuclear quantum effects ( nqe ) , as provided by the feynman discretized path integral technique , is expected to further decrease the strength of the interaction between the cl@xmath1 ion and water as a result , a further increase in @xmath19 towards the experimental value is anticipated with such a theoretical treatment of this fundamental aqueous ionic solution . in this work , we have systematically studied the solvation and electronic structure of the aqueous chloride ion solution using dft - based aimd which incorporates exact exchange ( @xmath0 ) and non - local vdw / dispersion interactions in the underlying xc potential . from an analysis of @xmath2 and @xmath3 , we find that the inclusion of exact exchange and non - local vdw / dispersion interactions _ via _ the use of the pbe0+ts - vdw(sc ) xc functional effectively _ weakens _ the interactions between the cl@xmath1 ion and the surrounding water molecules residing in the first solvation shell and yields a mean cl - o coordination number , @xmath33 , that is in excellent agreement with the available experimental data . moreover , we found that most configurations at the vdw - inclusive hybrid pbe0+ts - vdw(sc ) level of theory can be characterized by 6-fold cl - o coordination and are predominantly comprised of distorted trigonal prism structures findings which are strongly indicative of a significantly disordered first solvation shell surrounding the cl@xmath1 ion . by performing a systematic series of band structure calculations on molecular structures ( configurations ) generated from aimd simulations with various differing xc potentials , we were able to selectively isolate the effects arising from the underlying molecular structure as well as the choice of the electronic structure method on the energy levels of the solvated cl@xmath1 ion relative to the vbm of liquid water ( @xmath19 ) . in doing so , we found that while the band structure of liquid water is rather insensitive to the underlying molecular configurations , the relative positions of the orbital levels associated with the solvated cl@xmath1 ion with respect to the vbm of liquid water strongly depend on the underlying molecular configurations . here , the effective weakening of the hydrogen bonds existing between the lone pair electrons of the cl@xmath1 ion and the water molecules residing in the first solvation shell due to the collective effects of @xmath0 and non - local vdw / dispersion interactions leads to a decrease in the degree of hybridization between the cl@xmath1 @xmath27 and the oxygen @xmath20 orbitals located at the edge of the liquid water valence band . in the same breath , the pdos of the solvated cl@xmath1 ion is also significantly affected by the choice of dft functional employed in the electronic structure calculation ; as the fraction of @xmath0 was increased from 25% ( pbe0 ) to 50% ( bhlyp ) , @xmath19 steadily increased from 0.34 ev to 0.92 ev , thereby rapidly approaching the experimental value of 1.251.50 ev . as a final set of remarks , we note that the quantitative description of the microscopic structure of the aqueous cl@xmath1 ion solution may change by further refinement of the underlying xc functional approximation , which may be accomplished by reducing the self - interaction error _ via _ fine - tuning the exchange correction @xcite , and/or including a better description of vdw / dispersion interactions _ via _ the inclusion of beyond - pairwise interactions , as provided , for example , by the recently proposed many - body dispersion ( mbd ) scheme @xcite . as mentioned above , structural refinements are also expected to result from a proper treatment of nuclear quantum effects _ via _ the feynman discretized path - integral approach , which is expected to further weaken the interactions between the solvated cl@xmath1 ion and its aqueous surroundings ; the use of such configurations in conjunction with a higher - level electronic structure theory , such as hedin s gw self - energy approximation @xcite and/or the random phase approximation ( rpa ) @xcite , should lead to even better agreement with pes experiments on the relative energetics among the solvated cl@xmath1 orbitals and the vbm of liquid water . and m.k are grateful for partial financial support from the u.s . department of energy , office of basic energy sciences , scidac award de - fg02 - 12er16333 and computational support from xsede grant no . x.w . acknowledges support from the american chemical society petroleum research fund ( acs prf ) under grant no . 53482-dni6 . r.d . and b.s . acknowledge support from the scientific discovery through advanced computing ( scidac ) program through the department of energy ( doe ) under grant nos . de - sc0005180 and de - sc0008626 . this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s . department of energy under contract no . de - ac02 - 05ch11231 . this research used resources of the argonne leadership computing facility at argonne national laboratory , which is supported by the office of science of the u.s . department of energy under contract de - ac02 - 06ch11357 . p. giannozzi , s. baroni , n. bonini , m. calandra , r. car , c. cavazzoni , d. ceresoli , g. l. chiarotti , m. cococcioni , i. dabo , a. dal corso , s. de gironcoli , s. fabris , g. fratesi , r. gebauer , u. gerstmann , c. gougoussis , a. kokalj , m. lazzeri , l. martin - samos , n. marzari , f. mauri , r. mazzarello , s. paolini , a. pasquarello , l. paulatto , c. sbraccia , s. scandolo , g. sclauzero , a. p. seitsonen , a. smogunov , p. umari and r. m. wentzcovitch , j. phys . : condens . matter * 21 * , 395502 ( 2009 ) .	in this work , the solvation and electronic structure of the aqueous chloride ion solution was investigated using density functional theory ( dft ) based _ ab initio _ molecular dynamics ( aimd ) . from an analysis of radial distribution functions , coordination numbers , and solvation structures , we found that exact exchange ( @xmath0 ) and non - local van der waals ( vdw ) interactions effectively _ weaken _ the interactions between the cl@xmath1 ion and the first solvation shell . with a cl - o coordination number in excellent agreement with experiment , we found that most configurations generated with vdw - inclusive hybrid dft exhibit 6-fold coordinated distorted trigonal prism structures , which is indicative of a significantly disordered first solvation shell . by performing a series of band structure calculations on configurations generated from aimd simulations with varying dft potentials , we found that the solvated ion orbital energy levels ( unlike the band structure of liquid water ) strongly depend on the underlying molecular structures . in addition , these orbital energy levels were also significantly affected by the dft functional employed for the electronic structure ; as the fraction of @xmath0 was increased , the gap between the highest occupied molecular orbital of cl@xmath1 and the valence band maximum of liquid water steadily increased towards the experimental value .
the observation of the condensation of exciton - polaritons @xcite has created a large amount of interest in the last decade @xcite . exciton - polariton condensates display fascinating properties such as superfluidity @xcite , vortex formation @xcite , and has been suggested for use in future technologies such as polaritronics , the polariton analogue of atomtronics @xcite , quantum simulators @xcite , and novel light sources @xcite . one recent development is the observation of dipolaritons bosonic quasiparticles formed in coupled double quantum wells embedded into a microcavity formed by two distributed bragg reflectors ( fig . 1 ) @xcite . compared to the exciton - polariton , which is a quasiparticle consisting of a superposition of a photon and an exciton , a dipolariton is a three - way superposition of a microcavity photon , direct exciton ( dx ) , and an indirect exciton ( ix ) . the dx is a bound electron - hole pair in the same quantum well , and an ix is a bound electron - hole pair between the quantum wells . the coupling to photons means that it is a new type of polariton , which shares similar properties to the exciton - polariton such as light effective mass , but in addition has a dipole moment @xcite . the dipole moment is expected to enhance the dipolariton - dipolariton interactions , but in a tunable way by varying the relative proportions of the photon , dx , and ix . this is interesting from a quantum optoelectronic standpoint with potential applications to coherent transfer between photons to electrons @xcite . although condensation of dipolaritons has not been observed to date , the light effective mass of the dipolaritons suggest that the prospect of this is rather promising . despite the experimental interest , currently there is no rigorous theoretical treatment of the properties of the dipolariton , in particular the effective dipolariton - dipolariton interaction . several works have in the past have calculated the effective exciton - exciton interaction @xcite , from which the polariton - polariton interaction can be obtained simply by multiplying by the exciton fraction . the exciton - exciton interaction originate from the coulomb interaction of the underlying electron and holes making up the excitons . the interaction is typically described as the sum of two contributions the `` direct '' and `` exchange '' contributions @xcite . the direct contribution corresponds to exciton - exciton scattering process @xmath0 , where @xmath1 denotes an exciton containing an electron @xmath2 and a hole @xmath3 . the dashed and undashed labels refer to wavefunction coordinate labels in first quantized formalism , where the total wavefunction is antisymmetrized with respect to the electrons and holes exchange . the exchange contribution corresponds to the exchange exciton - exciton scattering @xmath4 , where one of the underlying fermions is exchanged . it is well - known that under typical densities the dominant process for dxs is the exchange contribution , giving the standard interaction @xmath5 , where @xmath6 is the bohr radius of the exciton , @xmath7 is the sample area , @xmath2 is the charge of the electron , and @xmath8 is the permittivity of the semiconductor . in ref . @xcite , this procedure was generalized to ixs . here it was found that for a non - zero quantum well separation ( see fig . [ fig : device ] ) , both the direct and exchange contributions need to be taken into account to obtain the effective ix - ix interaction . in this paper we present a theoretical description of dipolaritons within coupled double quantum wells in semiconductor microcavity structures . in particular we give a detailed derivation of the effective dipolariton - dipolariton interaction . due to the three - way superposition , the interaction will result from the total of dx - dx interaction , ix - ix interaction , and the ix - dx interaction . while the dx - dx and ix - ix interaction were analyzed before , here we give a detailed calculation of the ix - dx interaction , which has not appeared in the literature before . each of the three contributions to the interactions have a direct and exchange contribution , and we will show that it is important to include exchange effects for the ix - dx interaction . after performing the full calculation , we are able to obtain a simple expression for the dipolariton - dipolariton interaction valid for low dipolariton densities . for readers that are uninterested in the details , the primary result of this paper ( [ mainresult ] ) , which gives the effective dipolariton - dipolariton interaction . as mentioned above , due to the light dipolariton mass , they are promising for realization of condensation . we give simple formulas for parameters that would describe the condensate , modeled in terms of a dissipative gross - pitaevskii equation @xcite , which is the standard way to describe polariton condensates . we give estimates of parameters and show the suitability of dipolaritons for condensation . this paper is organized as follows . in sec . [ sec : dipolham ] we derive the dipolariton hamiltonian starting from the photon , direct exciton , and indirect exciton constituents . this will serve to identify what quantities are necessary to calculate regarding the various interaction contributions between the constituent species . in sec . [ sec : eff ] we discuss the dx - dx , ix - ix , and ix - dx interactions . the ix - dx interaction is calculated in detail while the known results for the dx - dx and ix - ix interaction are quoted for convenience . in sec . [ sec : gpequation ] we present the dissipative gross - pitaevskii equation for dipolaritons , along with simple equations for the parameters and numerical estimates . finally , the summary of the results and the conclusions follow in sec . [ sec : conclusions ] . a typical semiconductor microcavity system for dipolaritons is shown in fig . [ fig : device ] . a coupled double quantum well is located between two sets of distributed bragg reflectors ( dbrs ) , forming a microcavity . as is the case with standard exciton - polaritons , the microcavity allows for strong coupling between direct excitons in one of the quantum wells . in addition to this , the barrier between the two quantum wells is made sufficiently thin such that tunneling may occur between them . an applied bias voltage in the @xmath9-direction ensures that only the electron has the possibility of tunneling into the other quantum well . due to the small effective mass of the electron , it may tunnel between the two quantum wells . on the other hand , the tunneling of the hole is negligible because of its larger effective mass and energy separation of hole levels in a coupled double quantum well @xcite . thus there is a significant probability that the direct exciton may turn into an indirect exciton due to electron tunneling . according to the parameters in ref . @xcite , the tunneling amplitude may be tunable to the same order of the rabi splitting , thus a coherent superposition of the photon , direct exciton , and indirect exciton is a good approximation . denoting the bosonic annihilation operators of the photon , dx , and ix as @xmath10 respectively , the total hamiltonian of the system can be written @xmath11 the kinetic energy of the cavity photon , dx , and ix is @xmath12 where @xmath13 the zero energy point is taken to be the energy of the zero momentum @xmath14 mode of the direct excitons . the @xmath14 modes of the photon and indirect exciton are taken to have a detuning of @xmath15 and @xmath16 , respectively . the parameters involved in the above hamiltonian are the exciton mass @xmath17 , where @xmath18 ( @xmath19 ) is the effective electron ( hole ) mass , and the photon effective mass is @xmath20 . @xmath21 is the two dimensional center of mass position of the respective particles . the remaining single - particle terms are the rabi coupling between the direct excitons and photons , @xmath22 \end{aligned}\ ] ] and the tunneling between the direct excitons and indirect excitons @xmath23 .\end{aligned}\ ] ] here @xmath24 is the rabi coupling photons and the dx , and the tunneling energy between dx and ix is @xmath25 . the fourier transforms are defined as @xmath26 . finally , the remaining term in ( [ grandham ] ) is @xmath27 which are the non - linear interaction terms arising from dx - dx scattering ( @xmath28 ) , ix - ix scattering ( @xmath29 ) , ix - dx scattering ( @xmath30 ) , and a so - called `` saturation interaction '' ( @xmath31 ) @xcite due to bosonization of the rabi coupling . @xmath32 . \label{satham}\end{aligned}\ ] ] we explain in more detail the origin of these terms and explicit expressions for the matrix elements @xmath33 in the following section . the non - interacting polariton hamiltonian ( [ polham ] ) may be diagonalized by the linear transformation @xmath34 where @xmath35 are hopfield coefficients for the photon , direct exciton , and indirect exciton components , respectively . the new quasiparticles are lower polariton ( lp ) , middle polariton ( mp ) , and upper polaritons ( up ) . the diagonalized non - interacting dipolariton hamiltonian is @xmath36 , \end{aligned}\ ] ] where @xmath37 are the energy eigenvalues of the single particle hamiltonian @xmath38 a typical plot of the lp , mp , up dispersions are shown in fig . [ fig : disp ] . although the three - way superposition makes the understanding of the dispersion more complicated than the simple anticrossing picture for exciton - polaritons , there is a simple way to understand the qualitative features of the spectrum . first consider the dx and ix alone ( i.e. @xmath39 ) , and notice that due to the equality of the dx and ix mass the dispersions are separated by a constant amount for all @xmath40 . for @xmath41 the two quasiparticles are @xmath42 with energies @xmath43 . now reinstating the photon coupling , we may think of the dipolariton as being a further admixture of the photon and the hybrid ix - dx particle . for the case shown in fig . [ fig : disp ] , the lp dispersion is pushed down due to the anticrossing of the photon dispersion with respect to the @xmath44 particle at energy @xmath45 . this creates a typical lp dispersion similar to exciton - polaritons , but offset in energy by @xmath46 . thus , as far as the lp dispersion is concerned , the dipolariton dispersion shows the same essential features as standard exciton - polaritons . for sufficiently low temperatures @xmath47 we may expect that only the lp branch is populated , and we may ignore the mp and up branches completely . dropping the `` lp '' labels in ( [ transmatrix ] ) , we may write an effective hamiltonian only for the lower polaritons @xmath48 where the effective lp interaction is @xmath49 we see that the effective lp interaction consists of four parts , the mutual dx scattering , mutual ix scattering , ix - dx scattering , and the saturation interaction . these terms are derived explicitly in the following section . in this section we present the effective polariton interaction that consists from four terms : direct exciton - direct exciton ( dx - dx ) , indirect exciton - indirect exciton ( ix - ix ) , indirect exciton - direct exciton ( ix - dx ) interactions and the saturation interaction . calculation of the dx - dx and ix - ix interaction has already been performed in several previous works , hence we give a brief overview and restate the main results . for the dx - dx interaction , an effective hamiltonian in terms of excitons is derived from an electron - hole hamiltonian involving coulomb interactions @xcite . the ix - ix interaction has been calculated in ref . @xcite by generalizing the methods of refs . while a variety of methods exist to derive the effective hamiltonian , we follow the methods of de - leon and laikhtman @xcite which gives a transparent and systematic way of obtaining the relevant quantities . we summarize the approach for the ix - ix interaction , which reduces to the dx - dx interaction by setting the interwell distance @xmath50 to zero . the method starts with an antisymmetrized two ix wavefunction @xmath51 - \frac{1}{\sqrt{2 } } \big [ \psi_{\bm{q } } ( \bm{r}_{e ' } , \bm{r}_h ) \psi_{\bm{q } ' } ( \bm{r}_e , \bm{r}_{h ' } ) \nonumber \\ & + \psi_{\bm{q } } ( \bm{r}_e , \bm{r}_{h ' } ) \psi_{\bm{q}'}(\bm{r}_{e ' } , \bm{r}_h ) \big ] \big ) \label{antisym}\end{aligned}\ ] ] where @xmath52 is an ix wavefunction with center of mass momentum @xmath53 , taken as the two dimensional 1s wavefunction @xcite in the quantum well plane and a delta - function in the @xmath54-direction for the electrons and holes . this assumes that for an ix the electron is always perfectly localized in the electron quantum well , and similarly for the holes . here @xmath55 are the three dimensional coordinates of the electrons and holes respectively . the hamiltonian of the two exciton system is @xmath56 with @xmath57 ( @xmath58 is the permittivity in gaas , where @xmath59 is the permittivity in free space ) . taking the expectation value of ( [ twoexcham ] ) with respect to @xmath60 and @xmath61 one obtains the effective interaction @xcite @xmath62 \label{totalinteractionfull}\end{aligned}\ ] ] where @xmath63 is the 2d bohr radius , @xmath7 is the trapping area of the excitons , and @xmath64 is the reduced mass @xmath65 . the dimensionless integrals @xmath66 and @xmath67 are given in eqs . ( a2 ) and ( a5 ) and plotted in figs . 2 and 3 of ref . @xcite respectively . the dx - dx interaction is simply the same as this but evaluated at @xmath68 @xmath69 expressions for @xmath66 and @xmath67 are given in eqs . ( 20 ) and ( b1 ) and plotted in figs . 1 and 2 of ref . @xcite respectively . as discussed in ref . @xcite , the most relevant momentum scale in ( [ totalinteractionfull ] ) and ( [ dxdxint ] ) is the @xmath70 limit since their characteristic momentum scale is of the order of @xmath71 which is quite large compared to typical experimental situations . from fig . 4 in ref . @xcite to a good approximation the interaction is . therefore these terms do not need to be included in eq . ( 28 ) and ( 40 ) , resulting in an extra factor of 2 . ] @xmath72 setting @xmath73 agrees with the standard estimate of @xmath74 for the dx - dx interaction . similar methods may be used to derive the `` saturation '' interaction , originating from corrections to the rabi coupling due to the antisymmetrized two exciton wavefunction ( [ antisym ] ) . following ref . @xcite we have @xmath75 where @xmath76 is given in eq . ( a8 ) and fig . 3 in ref . this factor again has a characteristic momentum scale of @xmath71 , thus for typical experiments only the @xmath70 is of significance . we will therefore be interested in the value @xmath77 the ix - dx interaction is calculated using analogous methods as described in sec . [ sec : interaction ] . we start with a composite ix and dx wavefunction written as @xmath78 . \label{totalwave}\end{aligned}\ ] ] here only the coordinates for the holes have been antisymmetrized as the electrons lie in different layers and are distinguishable . in ( [ totalwave ] ) @xmath79 denotes the ix wavefunction , and @xmath80 is the dx wavefunction each with center of mass momentum @xmath53 . taking the matrix element of ( [ totham ] ) with @xmath81 and @xmath82 we obtain the contributions @xmath83 where @xmath84 the last term in ( [ totalixdx ] ) is a correction term to take into account for the fact that the wavefunction ( [ totalwave ] ) does not obey orthonormality , and gives spurious `` kinematic corrections '' @xcite . the correction factor is @xmath85 with the non - orthonormality factor @xmath86 to evaluate the expressions above , we use an approximate form for the ix ground state wavefunction , as obtained in ref . @xcite : @xmath87 where @xmath88 , @xmath89 , @xmath90 , and @xmath91 . in eq . ( [ approximateexciton ] ) , the exciton is considered to be trapped in a large area @xmath7 , such that the center of mass wavefunction is of the form of a plane wave . we have assumed that in the @xmath54 direction the electrons ( holes ) are completely confined to their respective quantum wells at position @xmath92 ( @xmath93 ) , with delta function wavefunctions for simplicity . the wavefunction that describes the state of the exciton for the relative coordinates is @xmath94 , \end{aligned}\ ] ] where @xmath95 and @xmath96 is a normalization factor . the dx wavefunction can be obtained with the substitution @xmath97 , giving @xmath98 the evaluations of the various terms are deferred to the appendix [ app : effectiveinteraction ] . finally we find the direct term to be @xmath99\delta ( \bm{q } ) \nonumber \\ & + \frac{1}{a } \frac{e^2}{4 \pi \epsilon } a_b \left ( \frac{2}{\pi } \right)^2 i_{\mbox{\tiny dir } } ( q , d ) .\end{aligned}\ ] ] where @xmath100 is the binding energy of a @xmath101 exciton for the ix and dx , respectively . the function @xmath102 is plotted for various @xmath50 in fig . [ fig : direct ] . as is the case for the dx - dx interactions , in the limit of @xmath103 this contribution becomes zero for all @xmath50 . thus the dominant contribution to the ix - dx interaction is due to the exchange term . we evaluate the exchange term to be @xmath104 { \cal a}(\bm{q } , \bm{q}',\bm{q } ) \nonumber \\ & - \frac{1}{a } \frac{e^2}{4 \pi \epsilon } a_b \left ( \frac{2}{\pi } \right)^2 i_{\mbox{\tiny exch } } ( \delta q , q,\theta,\beta_e , d ) , \end{aligned}\ ] ] where @xmath105 and @xmath106 is the angle between @xmath107 and @xmath108 . the first term proportional to @xmath109 cancel with the corrections due to non - orthonormality ( i.e. the last term in ( [ totalixdx ] ) ) . numerical evaluations of the exchange integral @xmath67 are shown in fig . [ fig : exchange ] . we see that this term survives in the limit @xmath103 and to a good approximation will give all of the contribution to the ix - dx scattering . substituting ( [ finaldirect ] ) and ( [ holeexchange ] ) into ( [ totalixdx ] ) we obtain the final effective hamiltonian for the two - exciton system . subtracting the kinetic energy and binding energy terms , we obtain an expression for exciton - exciton interaction @xmath110\end{aligned}\ ] ] as before , the primary region of experimental interest is the @xmath103 limit . to a good approximation the dimensionless integrals obey in this limit @xmath111 a comparison of the approximation ( [ approxiexch ] ) with a numerical integration is shown in fig . [ fig : exchange ] . we see that the approximation works rather well for the full range , and falls off fairly quickly for interwell distances of the order of the bohr radius . our final expression is therefore @xmath112 ( [ dipolham ] ) gives an effective hamiltonian that describes the low energy excitations of the dipolariton system . the effective dipolariton - dipolariton interaction may be obtained by substituting the results of sec . [ sec : eff ] into ( [ ulpexpression ] ) . as with exciton - polaritons , what is of primary interest is the condensation of these particles . we now describe the relevant parameters so that an effective theory of condensation of dipolaritons may be written . first let us evaluate the basic quantities of the dipolariton lp mass and lifetime . this may be calculated by writing ( [ singleparticleham ] ) as @xmath113 where the @xmath114 contains the decay rates of each of the components related to the lifetimes by @xmath115 . first treating @xmath114 as a perturbation to @xmath116 , we obtain the decay rate of the lps @xmath117 as with exciton - polaritons , for the case that @xmath118 we have @xmath119 . thus the dipolariton lp lifetime is of the order of the photon lifetime . the lp mass is obtained by treating @xmath120 as a perturbation to @xmath121 , and ignoring @xmath114 for simplicity . the lp mass is @xmath122 where for @xmath123 we have omitted the momentum labels on the hopfield coefficients for brevity . for the typical case where @xmath124 , we have @xmath125 . again , the dipolariton lp mass is of the order of the photon effective mass . in the case of zero detuning @xmath126 , the coefficients in ( [ effectivemasseq ] ) are @xmath127 and @xmath128 . finally , for condensation of dipolaritons , the relevant interaction parameter is the low energy scattering @xmath70 . compiling the results of sec . [ sec : eff ] and substituting this into ( [ ulpexpression ] ) , we obtain @xmath129 \nonumber \\ & + 3.5 \hbar \omega \frac{a_b^2}{a } \|x\|^2 ( c^ * x + x c^ * ) . \label{mainresult}\end{aligned}\ ] ] due to the relatively weak ix - dx interaction , most of the contribution will result from the ix - ix and the dx - dx interactions . thus the dipolariton - dipolariton interactions are generally of the same order as those for standard exciton - polaritons . we are now in a position to write down an equation which describes the dipolariton condensate . due to the close similarity of the physics of dipolaritons to exciton - polaritons , we may assume that condensation of dipolaritons also occurs to form a macroscopically occupied ground state @xcite . for details of exciton - polariton condensation see review articles such as refs . incoherent pumping of the dipolariton system results in initially a large population of reservoir excitons , corresponding to dx or ix depending on the pumping scheme . these excitons cool within the semiconductor via phonon emission , up to a bottleneck momentum , where the photon fraction becomes appreciable . a bottleneck population at momentum such that @xmath130 is then created , after which dipolariton - dipolariton scattering becomes the dominant mechanism of dipolariton momentum transfer . from the results of sec . [ sec : eff ] , the magnitude of the dipolariton - dipolariton scattering is of the same order as for standard excitons , hence this should occur rather efficiently . at sufficiently high reservoir densities , a macroscopic population of dipolaritons should form at @xmath131 . for non - zero temperatures , @xmath132 , there is no condensate in an infinite two - dimensional system . however , condensation is possible in a finite - sized system @xcite . a trapped , ideal bose gas undergoes a transition to a condensed state at the critical temperature @xmath133 , where @xmath134 is the spin degeneracy , @xmath135 is the polariton density in a trap @xcite . for lower polaritons with density @xmath136 @xmath137 , one obtains @xmath138 k. for exciton - polaritons , the dissipative gross - pitaevskii ( gp ) equation captures the condensate dynamics to a good approximation @xcite . for low enough temperatures below @xmath139 we may thus also write for dipolaritons @xmath140 \nonumber \\ & + g \| \varphi ( \bm{r } ) \|^2 + 2 g n ( \bm{r } ) \big ] \varphi ( \bm{r})\end{aligned}\ ] ] and the reservoir obeys @xmath141 here @xmath142 is the macroscopic wavefunction of a dipolariton condensate , @xmath143 is the spatial trapping potential , @xmath144 is the reservoir density , @xmath145 is the pumping rate of the reservoir , @xmath146 is the decay rate of the reservoir , and @xmath147 is the stimulated scattering of the reservoir excitons into the @xmath131 dipolariton mode . the only remaining unspecified parameters in the dissipative gp - equation are the reservoir - condensate scattering @xmath148 and the pumping rate @xmath145 . even for the exciton - polariton case these two parameters are typically put in phenomenologically , due to the difficulty of precisely modeling the reservoir . however , we do know that the scattering occurs due to interactions of an incoherent reservoir of dx and ix , which was calculated in sec . [ sec : eff ] to be very similar to the dx scattering for the exciton - polariton case @xcite . therefore it is reasonable to assume similar values to that used for exciton - polaritons , as with the phenomenological pump rate @xmath145 . we have obtained a simple theoretical description of dipolaritons in a coupled double well microcavity system . overall , the effective parameters as derived in sec . [ sec : gpequation ] suggest that the admixture of the indirect excitons give only a minor modification of the essential parameters . as with exciton - polaritons , the effective dipolariton mass is of the order of the light photon effective mass , and the lifetime is of the order of the photon lifetime . the dipolariton lp dispersion shows the same general behavior with a sharp anticrossing at momenta @xmath130 . the effective dipolariton - dipolariton interaction was calculated and was found to be expressible by the simple relation eq . ( [ mainresult ] ) . there are four contributions to the effective interaction , resulting from the dx - dx , ix - ix , ix - dx scattering , and the saturation interaction . the relative strength of these contributions depend upon the photon , dx , and ix fractions , which may be tuned by changing the detuning and tunneling strength @xmath149 . under typical parameters where each of the fractions are comparable , the dominant effective interaction originates from the ix - ix interaction and the dx - dx interaction . due to the fast fall - off of the ix - dx contribution with the interwell distance @xmath50 ( fig . [ fig : exchange](b ) ) , this contribution is typically the weakest of the four . due to the dipolar nature of the dipolaritons , one may expect that the ix - ix would be considerably stronger than the dx - dx interactions , which originate from purely an exchange effect . however , as may be observed from ( [ ixixint ] ) the increase in interaction is relatively weak , only linearly increasing with the interwell separation @xmath50 . this would suggest that contrary to expectation , the range of tunability of the dipolaritons is only moderate compared to exciton - polaritons , which may also be tuned by varying the photon and exciton fractions . however , the similar parameters and dispersion characteristics suggest that there should be no impediment in principle for condensation of dipolaritons , with similar physics to exciton - polaritons . similar configurations may be possible with a polaritons formed by a coupled double graphene layer in a microcavity structure @xcite . even without the feature of tunability , this would open a fascinating variant of the exciton - polariton condensate both from a fundamental point of view and technological applications . this work is supported by the transdisciplinary research integration center , the okawa foundation , the inamori foundation , ntt , and jsps kakenhi grant number 26790061 . g.k . is grateful to professional staff congress city university of new york for support , award # 66140 - 00 44 . . ( [ dirdefinition ] ) may be evaluated by making a change of variables to @xmath150 and @xmath151 , after which we obtain @xmath152\delta ( \bm{q } ) \nonumber \\ & + \frac{e^2}{4 \pi \epsilon a } \int d^2 \rho d^2 \rho ' \frac{2 \pi}{q } \| g(\rho , d ) \|^2 \| g(\rho ' , 0 ) \|^2 \nonumber \\ & \times \big [ e^{-i \bm{q } \cdot \beta_h ( \bm{\rho } - \bm{\rho } ' ) } e^{-dq } + e^{i \bm{q } \cdot \beta_e ( \bm{\rho } - \bm{\rho } ' ) } \nonumber \\ & - e^{- i \bm{q } \cdot ( \beta_h \bm{\rho } + \beta_e \bm{\rho } ' ) } e^{-dq } - e^{i \bm{q } \cdot ( \beta_e \bm{\rho } + \beta_h \bm{\rho } ' ) } \big ] \end{aligned}\ ] ] eq . ( [ finaldirect ] ) may be obtained by performing the @xmath153 and @xmath154 integrals separately and using the rotational invariance of @xmath108 . [ fig : direct ] is obtained by evaluating @xmath155,\end{aligned}\ ] ] where @xmath156 and @xmath157 and @xmath158 is the bessel function of the first kind . @xmath159 + q a_b [ -x \cos \theta_x -\beta y_1 \cos \theta_1 + ( 1-\beta ) y_2 \cos \theta_2 ] \ } \nonumber \\ & \times g ( a_b \sqrt{(y_2 \cos \theta_2 - y_1 \cos \theta_1 - x \cos \theta_x)^2 + ( y_2 \sin \theta_2 - y_1 \sin \theta_1 - x \sin \theta_x)^2 } , 0 ) g ( a_b x , d ) g ( a_b y_1 , 0 ) g ( a_b y_2 , d ) \nonumber \\ & \big [ \frac{1}{\sqrt{y_1 ^ 2 + x^2 + 2 y_1 x \cos ( \theta_1 - \theta_x)+ ( d / a_b)^2 } } + \frac{1}{\sqrt{y_2 ^ 2 + x^2 - 2 y_2 x \cos ( \theta_2 - \theta_x ) } } -\frac{1}{\sqrt{y_2 ^ 2 + ( d / a_b)^2 } } -\frac{1}{y_1 } \big ] . \label{iexchexpression}\end{aligned}\ ] ]	dipolaritons are a three - way superposition of photon , a direct exciton , and an indirect exciton that are formed in coupled quantum well microcavities . as is the case with exciton - polaritons , dipolaritons have a self - interaction due to direct and exchange effects of the underlying electrons and holes . here we present a theoretical description of dipolaritons and derive simple formulas for their basic parameters . in particular , we derive the effective dipolariton - dipolariton interaction taking into account of exchange effects between the excitons . we obtain a simple relation to describe the effective interaction at low densities . we find that dipolaritons should condense under suitable conditions , described by a dissipative gross - pitaevskii equation . while the parameters for condensation are promising , we find that the level of tunability of the interactions is limited .
half - metallic ferromagnets @xcite offer a unique opportunity for studying the electronic states of strongly correlated electron systems . here , only the majority - spin electrons form the fermi surface with a gapped minority - spin band @xcite and can couple with excitations of the spin ( and in some cases orbital ) degrees of freedom of the system . the situation therefore should attract much interest , in particular , when the relevant electrons are strongly correlated , leading the system to double - exchange ferromagnetism @xcite and metal - insulator transition ( mit ) . in this paper , we will show that a chromium oxide k@xmath0cr@xmath1o@xmath2 with the hollandite - type crystal structure @xcite belongs to this class of materials and can provide a good opportunity for further development of the physics of strong electron correlations . the crystal structure of k@xmath0cr@xmath1o@xmath2 ( see fig . 1 ) belongs to a group of hollandite - type phases where one - dimensional ( 1d ) double strings of edge - shared cro@xmath3 octahedra forms a cr@xmath4o@xmath2 framework of a tunnel structure , wherein k ions reside.@xcite cr ions are in the mixed - valent state of cr@xmath5 ( @xmath6 ) : cr@xmath7 ( @xmath8 ) @xmath9 , and hence with 2.25 electrons per cr ion . it has recently been reported @xcite that the phase transition from the pauli - paramagnetic metal to ferromagnetic metal occurs at @xmath10 k by lowering temperatures , where the ferromagnetic state has a full spin - polarization of 18 @xmath11 per formula unit ( f.u . ) at low temperatures , which is a realization of saturated ferromagnetism . in addition to this phase transition , it has also been reported @xcite that another phase transition occurs from the ferromagnetic metal to ferromagnetic insulator at @xmath12 k , suggesting that the charge gap opens below @xmath13 . surprisingly , the spin polarization is hardly affected by this mit and no structural distortions associated with this mit have been observed so far.@xcite the mechanism of the mit of this material has therefore been a great puzzle . + in this paper , we perform the first - principles electronic structure calculations based on the generalized gradient approximation ( gga ) in the density - functional theory ( dft ) and we predict that the materials @xmath14cr@xmath1o@xmath2 ( a = k and rb ) belong to a new class of half - metallic ferromagnets , i.e. , the majority - spin electrons are metallic , whereas the minority - spin electrons are semiconducting with a band gap . we also show from the gga and gga+@xmath15 calculations that the double - exchange mechanism is responsible for the observed saturated ferromagnetism . we then discuss possible mechanisms of the mit of this material and argue that the formation of an incommensurate , long - wavelength spin and charge density wave ( dw ) due to fermi - surface nesting may be the origin of the mit of this material . the opening of the gap in the majority - spin band should , however , be detrimental to the double - exchange ferromagnetism , for which we consider possible reconciliations . we hope that our results will encourage further experimental studies of this intriguing material . this paper is organized as follows : in sec . ii , we show our method of calculations . in sec . iii , we present our results of calculations and discuss the origin of ferromagnetism and possible mechanisms of the mit of k@xmath0cr@xmath1o@xmath2 . summary of our work and prospects for future studies are given in sec . the electronic structure calculation in gga is carried out by employing the computer code wien2k,@xcite which is based on the full - potential linearized augmented - plane - wave ( flapw ) method . we use the exchange - correlation potential of ref.@xcite . the spin polarization is allowed . the spin - orbit interaction is not taken into account . we use 1,221 @xmath16 points in the irreducible part of the brillouin zone in the self - consistent calculations . we use the plane - wave cutoff of @xmath17 bohr@xmath18 . the gga+@xmath15 calculation @xcite is also made to see the effects of on - site electron correlation @xmath15 on the band structure . we assume the experimental crystal structure of k@xmath0cr@xmath1o@xmath2 observed at room temperature with the lattice constants of @xmath19 and @xmath20 .@xcite the bravais lattice is body - centered tetragonal and the primitive unit cell ( u.c . ) contains four cr ions , one k ion , and eight o ions , i.e. , kcr@xmath21o@xmath1 . the local structure around cr ions is shown in fig . 2 . we use the code xcrysden@xcite for graphical purposes . the total ground - state energies obtained in gga are @xmath22 ryd / u.c . in the ferromagnetic state and @xmath23 ryd / u.c . in the paramagnetic state , resulting in the energy gain of @xmath24 ev / u.c . due to stabilization by the spin polarization , which is realized already in the gga level . in the ground state , we have the full spin polarization with the magnetic moment of @xmath25 @xmath11/u.c . , which consists of the contributions from cr ions , 2.043 @xmath11/atom , o(1 ) ions , 0.014 @xmath11/atom , and o(2 ) ions , @xmath26 @xmath11/atom , in the atomic spheres , and @xmath27 @xmath11/u.c . , in the interstitial region . note that the o(2 ) ions inside the double strings of cr ions have a negative spin polarization , whereas the o(1 ) ions connecting the double strings have a positive spin polarization . the calculated density of states ( dos ) is shown in fig . 3 in a wide energy range covering over the o @xmath28 and cr @xmath29 bands . we find three separate peaks in both the majority and minority spin bands : in the majority ( minority ) spin band , the o @xmath28 weight is located mainly at @xmath30 ( @xmath31 ) ev , cr @xmath29 weight with the @xmath32 symmetry at @xmath33 ( @xmath34 ) ev , and cr @xmath29 weight with the @xmath35 symmetry at @xmath36 ( @xmath37 ) ev . the hybridization between the o @xmath28 and cr @xmath29 bands is significantly large . the fermi level is located at a deep valley of the @xmath32 majority - spin band while it is located in the energy gap between the o @xmath28 and cr @xmath29 @xmath32 bands in the minority - spin band . thus , the half metallicity of this material is evident in the calculated dos . + the calculated orbital - decomposed partial dos , @xmath38 ( @xmath39 ) , in the cr @xmath29 @xmath32 region are shown in fig . 4 , where the two components are exactly degenerate , @xmath40 , for both paramagnetic and ferromagnetic states . the three @xmath32 orbitals are almost equally occupied by electrons in the paramagnetic state . in the ferromagnetic state , however , the @xmath41 orbitals is almost fully occupied by electrons and therefore holes are only in the @xmath42 and @xmath43 orbitals . also , the @xmath41 component has a rather high peak - like structure at @xmath440.7 ev below the fermi level , indicating an essentially localized character of the @xmath41 electrons . the @xmath42 and @xmath43 components , on the other hand , have a rather wide band spreading over @xmath45 ev around the fermi level , indicating an itinerant character of the @xmath42 and @xmath43 electrons . here , the admixture of the @xmath46 state of o(2 ) is significantly large : it occurs between the @xmath46 orbitals of o(2 ) and the @xmath42 and @xmath43 orbitals of its neighboring cr ions ( see fig . 2 ) . these results suggest that the double - exchange mechanism @xcite should work for the occurrence of ferromagnetism , as will be discussed further below . + + the situation may be clarified further if one observes the calculated band dispersion near the fermi level . the result is shown in fig . we find that a rather dispersionless narrow band of predominantly @xmath41 character is located at @xmath440.7 ev below the fermi level , extending over a large region of the brillouin zone . on the other hand , the dispersive @xmath32 bands of predominantly @xmath42 and @xmath43 character with strong admixture of the @xmath46 state of o(2 ) are located at @xmath45 ev and cross the fermi level . thus , we have the dualistic situation where the essentially localized @xmath41 electrons at @xmath440.7 ev below the fermi level interact with the itinerant @xmath42 and @xmath43 electrons of the bandwidth comparable with the intraatomic exchange energy of @xmath441 ev , whereby the hund s rule coupling gives rise to the ferromagnetic spin polarization via the double - exchange mechanism.@xcite to support this further , we make the gga+@xmath15 calculation for the present material ( of which the results are not shown here ) . we find that , as @xmath15 increases , the @xmath41 band shifts further away from the fermi level , leaving essentially no weight above the fermi level , whereas the @xmath42 and @xmath43 bands with strong admixture of the o(2 ) @xmath46 states are much less affected by the presence of @xmath15 . these results are consistent with what is expected in the double - exchange mechanism of ferromagnetism . we point out that the situation is very similar to the case of cro@xmath0 , which has so far been discussed in detail.@xcite another aspect noted in fig . 5 is that we have the semi - metallic band structure in the sense that the number of electrons is equal to the number of holes near the fermi level ( also see fig . 6 below ) ; there are 12 bands for the @xmath32 orbitals in the unit cell , which are fully spin - polarized and are occupied by 9 `` spinless fermions '' , i.e. , by 9 up - spin ( or majority - spin ) electrons . the integer filling of spinless fermions can give rise to either semiconducting or semimetallic band structure , of which the latter is realized in the present system . this result suggests that the mit observed in k@xmath0cr@xmath1o@xmath2 can possibly be a semimetal - to - semiconductor transition , where the band gap opens between the lower 9 bands and upper 3 bands due to any unknown reasons . to realize this , however , we need the symmetry reduction ( due to lattice distortion ) because the band degeneracy between the third and fourth bands ( counted from the top of the 12 @xmath32 bands ) at the p point of the brillouin zone can not be lifted without breaking the 4-fold rotational symmetry around the @xmath47 axis of the tetragonal lattice . without the lattice distortion , we could have a zero - gap semiconductor , but the experimental data on the temperature dependence of the electrical resistivity @xcite suggest that this possibility should be ruled out . also , because no indications of this type of lattice distortions at the mit have been observed so far,@xcite the semimetal - to - semiconductor transition seems to be unrealistic in k@xmath0cr@xmath1o@xmath2 . + + the calculated fermi surfaces of k@xmath0cr@xmath1o@xmath2 in the ferromagnetic state are shown in fig . there are 12 @xmath32 bands , 3 of which cross the fermi level and form the semimetallic fermi surfaces ; i.e. , the second and third bands ( counted from the top ) form the electron fermi surfaces ( see figs . 6 ( b ) and ( c ) ) and the fourth band forms the hole fermi surface ( see fig . 6(a ) ) . the wave functions at the fermi surfaces have predominantly @xmath42 and @xmath43 character with large admixture of the o(2 ) @xmath46 states as shown above . we find in fig . 6(a ) that there is a pair of the 1d - like parallel fermi surfaces , which are seen to have a very good nesting feature . the nesting vector is aligned roughly along the @xmath48-k@xmath49 direction and has the value @xmath50 or @xmath51 ( a deviation in the @xmath52 component will be discussed below ) . thus , the fermi - surface instability corresponding to the wavenumber @xmath53 , leading to formation of the incommensurate , long - wavelength ( with a period of @xmath447c in the real space ) dw , may be relevant with the opening of the charge gap in the present material . note that the spin and charge dws occur simultaneously with the same wavenumber @xmath53 since we have only the up - spin electrons . + to confirm the nesting features more precisely , we calculate the generalized susceptibility defined as @xmath54,@xcite where @xmath55 is the band dispersion and @xmath56 is the fermi distribution function at @xmath57 k. the calculated result is shown in fig . 7 , where the contribution from only the 61st band is given . contributions from other bands including interband contributions are rather monotone functions of @xmath16 over the entire brillouin zone . thus , the peak structure coming from the 61st band remains as a maximum even in the total susceptibility estimated in the constant matrix - element approximation . to be more precise , the peak structure at @xmath58 and @xmath59 seen in fig . 7 remains strong , irrespective of the value of @xmath52 , although there is a small variation in the @xmath52 plane . the true maximum appears at @xmath60 , or around @xmath61 , splitting and deviating slightly from @xmath61 in the @xmath52 plane . thus , if we include the effects of electron correlations , the susceptibility can diverge at this momentum @xmath53 , resulting in the formation of the incommensurate , long - wavelength charge and spin dw , which we hope will be checked by experiment in near future . + now , let us discuss possible mechanisms of the mit observed in k@xmath0cr@xmath1o@xmath2 at @xmath62 k. there are two types of scenarios : ( i ) correlation scenario where the electron correlations play an essential role and ( ii ) band scenario where the band structure solely determines whether the system is metallic or insulating . as a correlation scenario , we may have a possibility of the formation of the incommensurate , long - wavelength charge and spin dw due to the fermi - surface nesting as discussed above . this may give rise to the opening of the charge gap , resulting in the mit observed in the present material . as another correlation scenario , we may have the charge ordering ( co ) if the intersite coulombic repulsions between charge carriers are sufficiently strong . in the present system , there may occur the localization of 3 spinless @xmath32 holes per 4 cr site . we find that the possible spatial co patterns are equivalent to those of k@xmath0v@xmath1o@xmath2,@xcite which results in the doubling of the unit cell along the @xmath47 axis . as a band scenario , we may have the semimetal - to - semiconductor transition in the present system with the integer filling of spinless fermions as discussed in sec . iii c. here , no spatial localization of carriers occurs , but the band gap may open if the 4-fold rotational symmetry around the @xmath47 axis is broken due to lattice distortions and the distortion is sufficiently large . although the lattice distortion is essential for the mit in any of the above cases and its identification can specify which scenario is realized , no structural distortions associated with the metal - insulator transition have been observed so far.@xcite this seems to suggest that the simple distortions such as the doubling of the unit cell along the @xmath47 axis and the breaking of the 4-fold rotational symmetry around the @xmath47 axis may be ruled out . we therefore argue that the formation of the incommensurate , long - wavelength dw , the observation of which is sometimes not very easy , may be relevant with the mit in the present material . we hope that the anomaly at @xmath63 estimated above will be sought for carefully in future experiment . another experimental fact to be noted is that the spin polarization is hardly affected by the mit in this material . this is interesting because the double - exchange mechanism for ferromagnetism ( of which the energy gain occurs due to the first - order process of the hopping ) should cease to work if the charge gap opens clearly in the majority - spin band and coherent motion of carriers vanishes . the double - exchange ferromagnetism may be killed in such cases . however , if the charge gap is small enough , the second - order processes of the hopping of conduction electrons can give rise to the ferromagnetic interaction between localized spins as well and thus the ferromagnetism can be maintained . the vanishing first - order process seems to affect very little on the already fully spin - polarized system . experimentally,@xcite the electrical resistivity measurement shows that the temperature dependence is of a three - dimensional variable - range - hopping type rather than a thermal activation type and thus it is not clear whether the well - defined charge gap is present in this material . also , the charge gap expected in the formation of the incommensurate , long - wavelength dw state corresponding to @xmath53 may not be so large . the observed saturated ferromagnetism maintained under the gap formation seems to be related to such situations . we here point out that the present material k@xmath0cr@xmath1o@xmath2 is a doped negative charge - transfer - gap ( ct - gap ) type system@xcite in the zaanen - sawatzky - allen phase diagram.@xcite because k@xmath0cr@xmath1o@xmath2 is in the mixed - valent state , we need not invoke the negative ct - gap situation@xcite for metallization of the system , as is realized in cro@xmath0 where the metallization via self - doping mechanism operates.@xcite however , in a hypothetical k@xmath64cr@xmath1o@xmath2 system , where k ions are completely depleted and all the cr ions are in the @xmath6 state , this self - doping mechanism should be relevant for both metallization and double - exchange ferromagnetism unless the strong electron correlations leads the system to a mott - insulating ground state with spin @xmath65 local moments aligned antiferromagnetically or with local spin - singlet pairs of @xmath65 spins . we here point out that , even in k@xmath0cr@xmath1o@xmath2 , the negative ct - gap situation is realized between the o(2 ) @xmath28 and cr @xmath29 @xmath32 orbitals , as is evident in the calculated negative spin - polarization of the o(2 ) ions . in fact , results of our gga+@xmath15 calculation for k@xmath0cr@xmath1o@xmath2 indicate that the gap in the dos opens when @xmath15 is large enough although the fermi level is away from the gap due to the mixed - valent nature of the system . thus , the deficiency of k ions , if realized experimentally , can offer an important opportunity for studying the negative ct - gap situation and its carrier number dependence . in summary , we have made the first - principles electronic structure calculations and predicted that a chromium oxide k@xmath0cr@xmath1o@xmath2 of hollandite type should be a half - metallic ferromagnet . we have shown that the double - exchange mechanism is responsible for the observed saturated ferromagnetism . we have discussed the possible scenarios of the metal - insulator transition observed at low temperature , which include possibilities of the formation of the charge order and semimetal - to - semiconductor transition , but we have argued that the formation of the incommensurate , long - wavelength density wave of spinless fermions caused by the fermi - surface nesting may be the origin of the opening of the charge gap . we hope that these predictions will be checked by further experimental studies . before closing this paper , let us discuss some prospects for future studies of this material , which may include the following : + ( i ) this material offers an interesting opportunity for studying the effects of the majority - spin band gap on the electronic properties of half - metallic ferromagnets . in particular , the so - called non - quasiparticle states @xcite and the effects of the opening of the gap on the non - quasiparticle states should be pursued by the spectroscopic experiments such as photoemission and x - ray absorption experiment.@xcite + ( ii ) the transport properties of k@xmath0cr@xmath1o@xmath2 are also anomalous,@xcite as those of a `` bad metal'',@xcite which may also be interesting because not only the spin fluctuations but also the collective fluctuations of the orbital degrees of freedom between the degenerate @xmath42 and @xmath43 orbitals may play an important role . + ( iii ) introduction of the deficiency of k ions , if it could be made experimentally , would enable us to examine the effects of changing the doping rate on the gap formation , double exchange ferromagnetism , and half metallicity . we also point out that only the 1d - like fermi surface should remain by @xmath4410% hole doping if we assume the rigid - band shift of the fermi level ( see fig . 5 ) , suggesting the realization of the 1d electron system . + ( iv ) theoretically , one should study the three - band hubbard model with the @xmath41 , @xmath42 , and @xmath43 orbitals to see , in particular , the effects of orbital fluctuations on the electronic states of the system . the study may be extended to the @xmath66-@xmath67 model including the @xmath28 orbitals of o ions if one wants to take into account the negative ct - gap situation of the present system . techniques beyond the gga+@xmath15 method , such as the dynamical mean - field theory@xcite and variational cluster approach,@xcite are required to examine the presence of non - quasiparticle states and their deformation due to the gap formation . finally , we want to point out that , because the basic electronic states of the present system have many aspects in common with those of cro@xmath0 , it may be very useful to clarify similarities and differences in the electronic states between k@xmath0cr@xmath1o@xmath2 and cro@xmath0 in all the above respects . we would like to thank m. isobe , t. yamauchi , and y. ueda for informative discussions on experimental aspects of k@xmath0cr@xmath1o@xmath2 and s. ishihara and t. shirakawa for useful discussions on theoretical aspects . this work was supported in part by a grant - in - aid for scientific research ( no . 19014004 ) from the ministry of education , culture , sports , science and technology of japan . a part of computations was carried out at the research center for computational science , okazaki research facilities , and the institute for solid state physics , university of tokyo . m. i. katsnelson , v. y. irkhin , l. chioncel , a. i. lichtenstein , and r. a. de groot , rev . mod . phys . * 80 * , 315 ( 2008 ) . it is also possible that only the minority - spin electrons form the fermi surface with a gapped majority - spin band as in fe@xmath68o@xmath21 ; see a. yanase and k. siratori , j. phys . . jpn . * 53 * , 312 ( 1984 ) . c. zener , phys * 82 * , 403 ( 1951 ) . bull . * 11 * , 609 ( 1976 ) . h. okada _ et al . * 13 * , 1047 ( 1978 ) . o. tamada , n. yamamoto , t. mori , and t. endo , j. solid state chem . * 126 * , 1 ( 1996 ) . k. hasegawa , m. isobe , t. yamauchi , h. ueda , j. yamaura , h. goto , t. yagi , h. sato , and y. ueda , unpublished . p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , and j. luitz , _ wien2k , an augmented plane wave + local orbitals program for calculating crystal properties _ ( technische universitt wien , austria , 2002 ) ; http://www.wien2k.at . j. p. perdew , k. burke , and m. ernzerhof , phys . 77 * , 3865 ( 1996 ) . v. i. anisimov , i. v. solovyev , m. a. korotin , m. t. czyzyk , and g. a. sawatzky , phys . b * 48 * , 16929 ( 1993 ) . a. kokalj , comp . * 28 * , 155 ( 2003 ) . k .- h . schwarz , j. phys . f : metal phys . * 19 * , l211 ( 1986 ) . m. a. korotin , v. i. anisimov , d. i. khomskii , and g. a. sawatzky , phys . rev . lett . * 80 * , 4305 ( 1998 ) . a. yamasaki , l. chioncel , a. i. lichtenstein , and o. k. andersen , phys . b * 74 * , 024419 ( 2006 ) . m. isobe , s. koishi , n. kouno , j. yamaura , t. yamauchi , h. ueda , h. gotou , t. yagi , and y. ueda , j. phys . jpn . * 75 * , 73801 ( 2006 ) . s. horiuchi , t. shirakawa , and y. ohta , phys . b * 77 * , 155120 ( 2008 ) . m. sakamaki , s. horiuchi , t. konishi , and y. ohta , arxiv:0811.4338 . d. i. khomskii , preprint cond - mat/0101164 . j. zaanen , g. a. sawatzky , and j. w. allen , phys . rev . lett . * 55 * , 418 ( 1985 ) . j. rath and a. j. freeman , phys . b * 11 * , 2109 ( 1975 ) . d. m. edwards and j. a. hertz , j. phys . phys . * 3 * , 2191 ( 1973 ) . v. y. irkhin and m. i. katsnelson , phys . b * 73 * , 104429 ( 2006 ) . t. tsujioka , t. mizokawa , j. okamoto , a. fujimori , m. nohara , h. takagi , k. yamaura , and m. takano , phys . b * 56 * , r15509 ( 1997 ) . d. j. huang , h .- jeng , c. f. chang , g. y. guo , j. chen , w. p. wu , s. c. chung , s. g. shyu , c. c. wu , h .- j . lin , and c. t. chen , phys . b * 66 * , 174440 ( 2002 ) . d. j. huang , l. h. tjeng , j. chen , c. f. chang , w. p. wu , s. c. chung , a. tanaka , g. y. guo , h .- j . lin , s. g. shyu , c. c. wu , and c. t. chen , phys . b * 67 * , 214419 ( 2003 ) . e. z. kurmaev , a. moewes , s. m. butorin , m. i. katsnelson , l. d. finkelstein , j. nordgren , and p. m. tedrow , phys . b * 67 * , 155105 ( 2003 ) . l. ranno , a. barry , and j. m. d. coey , j. appl . phys . * 81 * , 5774 ( 1997 ) . k. suzuki and p. m. tedrow , phys . b * 58 * , 11597 ( 1998 ) . i. i. mazin , d. j. singh , and c. ambrosch - draxl , phys . b * 59 * , 411 ( 1999 ) . s. m. watts , s. wirth , s. von molnar , a. barry , and j. m. d. coey , phys . rev . b * 61 * , 9621 ( 2000 ) . l. craco , m. s. laad , and e. mller - hartmann , phys . lett . * 90 * , 237203 ( 2003 ) . l. chioncel , h. allmaier , e. arrigoni , a. yamasaki , m. daghofer , m. i. katsnelson , and a. i. lichtenstein , phys . b * 75 * , 140406(r ) ( 2007 ) .	based on the first - principles electronic structure calculations , we predict that a chromium oxide k@xmath0cr@xmath1o@xmath2 of hollandite type should be a half - metallic ferromagnet where the fermi level crosses only the majority - spin band , whereas the minority - spin band has a semiconducting gap . we show that the double - exchange mechanism is responsible for the observed saturated ferromagnetism . we discuss possible scenarios of the metal - insulator transition observed at low temperature and we argue that the formation of the incommensurate , long - wavelength density wave of spinless fermions caused by the fermi - surface nesting may be the origin of the opening of the charge gap .
the most massive galaxies in the universe occur in rich groups and clusters . these brightest cluster galaxies ( hereafter bcgs ) often reside close to the peak of the diffuse x - ray emission from the intracluster medium ( icm ; @xcite ) , and can have recessional velocities indistinguishable from the cluster mean ( @xcite ; but see also @xcite ) . their major axis is preferentially aligned with that of the host cluster , as defined by the projected distribution of cluster galaxies ( e.g. , @xcite ) or the icm x - ray emission @xcite , an effect which seems independent of cluster richness @xcite . furthermore , there is a clear relationship between the stellar luminosity of central bcgs and the total mass and x - ray luminosity of their host cluster @xcite . these observations suggest that many bcgs reside near the centers of the cluster gravitational potential , and that the formation of central bcgs is intimately linked to the formation and evolution of the cluster itself , as anticipated in a hierarchical structure formation scenario . this idea is supported by results of cosmological simulations which indicate that central bcgs are built up through a series of mergers near the cluster core ( e.g , @xcite ) . such simulations also suggest that these mergers must have been predominantly dissipationless , at least for bcg major mergers occurring at low redshift ( e.g. , @xcite ) . the notion that bcgs have a unique formation history is further reinforced by findings indicating that their properties are separate from those of other ( cluster ) galaxies . for example , the optical luminosity of bcgs is not drawn from the same distribution as that of other cluster members @xcite , bcgs tend to have spatially extended stellar halos not seen around other galaxies ( e.g. , @xcite ) , and they are more likely to be radio - loud than other galaxies of similar stellar mass @xcite . unfortunately , an observational understanding of the formation of bcgs is largely limited by the fact that few systems in the process of forming have actually been identified . multiple nuclei are common in bcgs at low and moderate redshifts , suggesting that galaxy galaxy mergers must be frequent ( e.g. , @xcite ) . in a few cases , multiple galaxies at the initial stages of merging have been identified at the centers of x - ray clusters @xcite , including one case in which tidal features provide direct evidence for galaxy galaxy interactions @xcite . however , as these examples all occur at moderate to high redshifts , detailed studies of the bcg formation process have not been possible . potential examples at lower redshift include known dumbbell galaxies @xcite , some of which constitute the brightest galaxy in x - ray clusters . however , at least some of these are likely to represent various stages in the mergers of _ cluster _ cores , each already containing a fully formed central dominant galaxy @xcite . as such , they may not offer a particularly clean view of the build - up of the bcg itself . for these reasons , it also remains uncertain to what extent bcg formation may have affected the thermodynamic history of baryons in x - ray bright cluster cores . for example , major galaxy galaxy mergers near the core of a cluster could provide a mechanism for distributing metals into the icm @xcite and so provide a partial explanation for the pronounced central metal excesses observed in many systems ( e.g. , @xcite ) . such mergers might also act to disrupt the formation of cool cores , and thereby help explain the puzzling dichotomy between `` cool - core '' and `` non cool - core '' groups and clusters ( e.g. , @xcite ) . here we present a _ study of a recently discovered low - redshift ( @xmath0 ) x - ray cluster , mz10451 , with a bcg in the process of forming . to the best of our knowledge , and if excluding dumbbell galaxies but including major merger candidates in low - redshift sdss groups and clusters @xcite , this represents the first known case of an ongoing central bcg major merger in an x - ray cluster in the nearby universe . the goal of the present investigation is two - fold . we aim to establish how the ongoing galaxy galaxy interaction is affecting the thermodynamics and chemical properties of the icm in the cluster core . the relatively low icm temperature of this system , @xmath1 kev , makes it attractive for a study of this kind , because any x - ray signatures of the interaction are likely to be more pronounced than in much more massive clusters . secondly , we also hope to shed light on the bcg formation process itself , e.g. by exploring whether the merging galaxies contain any hot ( or cold ) gas ( i.e. whether the merger is dissipationless ) , and whether they show evidence for strong nuclear activity induced by the interaction . we assume @xmath2 km s@xmath3 mpc@xmath3 , @xmath4 , and @xmath5 . the target redshift of @xmath6 then corresponds to a luminosity distance of @xmath7 mpc , and @xmath8 on the sky to 67 kpc . we further adopt solar abundances from @xcite , and a local galactic hi absorbing column of @xmath9 @xmath10 @xcite . unless otherwise stated , uncertainties are given at the 68% confidence level . as part of the _ xi _ groups survey @xcite , we have obtained comprehensive multi - wavelength data for a sample of 25 groups and poor clusters . we are using _ xmm - newton _ to study the x - ray properties of each system , and the imacs multi - object spectrograph @xcite on the 6.5-m baade / magellan telescope at las campanas to determine group membership down to very faint luminosities ( @xmath11 ) . in addition , we have obtained _ spitzer_/mips and _ galex _ uv imaging of each group field to probe the star formation properties of the group members . details of the x - ray , optical , and infrared data reduction can be found in @xcite and @xcite , while the _ galex _ analysis will be described in a forthcoming paper ( j. rasmussen et al . , in preparation ) . a summary of the x - ray , uv , and @xmath12 m data considered in the present paper is provided in table [ tab , log ] . lccc instrument & obs . i d & obs . date & @xmath13 + & & ( yyyy - mm - dd ) & + acis - s & 10467 & 2008 - 11 - 14 & 18.78 ks + _ chandra _ acis - s & 10800 & 2008 - 11 - 04 & 28.16 ks + _ chandra _ acis - s & 10802 & 2008 - 11 - 16 & 31.62 ks + _ xmm _ pn & 0305800901 & 2006 - 01 - 20 & 14.99 ks + _ xmm _ mos1 & & & 21.17 ks + _ xmm _ mos2 & & & 21.21 ks + _ galex _ & gi4_052009 & 2008 - 09 - 27 & 1607 s + _ spitzer _ mips & 22243840 & 2007 - 08 - 23 & 84 s + one of the most x - ray luminous systems in the _ xi _ survey so far is mz10451 . our optical spectroscopy of this system has identified 34 group members within the @xmath14 imacs field - of - view , with a velocity dispersion @xmath15 km s@xmath3 . a search of the 2df galaxy redshift catalog and the ned database reveals the presence of an additional 26 galaxies with concordant redshifts ( within @xmath16 of the group mean ) and within @xmath17 mpc from the x - ray peak , some of which may represent group members infalling for the first time . for these 60 system members , we derive a mean redshift of @xmath18 and a velocity dispersion @xmath19 km s@xmath3 @xcite . table [ tab , x_opt ] summarizes relevant x - ray and optical properties of the system derived here and in @xcite . remarkably , while our _ xmm _ data show symmetrically distributed hot gas out to several hundred kpc ( see section [ sec , results ] ) , consistent with a virialized system , the core of the system is dominated by _ two _ optically luminous early - type galaxies , as shown in figure [ fig , magellan ] . this central galaxy pair was previously noted by @xcite in their catalog of southern peculiar galaxies . in the following , the optically brighter northern galaxy in the pair will be referred to as galaxy a and its southern counterpart as galaxy b. lcc & value & source + @xmath20 & @xmath21 & magellan/2dfgrs + @xmath22 & 60 & magellan/2dfgrs + @xmath23 & @xmath24 km s@xmath3 & magellan/2dfgrs + @xmath25 & 430 kpc & _ xmm _ + @xmath26 & @xmath27 erg s@xmath3 & _ xmm _ + @xmath28 & @xmath29 kev & _ xmm _ + @xmath30 & @xmath31 kpc & _ xmm_/_chandra _ + @xmath32 & @xmath33 kpc & _ xmm_/_chandra _ + @xmath34 & @xmath35 kpc & _ xmm_/_chandra _ + @xmath36 & @xmath37 m@xmath38 & _ xmm_/_chandra _ + @xmath39 & @xmath40 m@xmath38 & _ xmm_/_chandra _ + @xmath41 & @xmath42 m@xmath38 & _ xmm_/_chandra _ + there is considerable evidence to suggest an ongoing merger between the two galaxies . firstly , their optical centers are separated by only @xmath43 ( @xmath44 kpc ) in projection , and their radial velocity offsets relative to the group mean are only 34 km s@xmath3 ( galaxy a ) and 2 km s@xmath3 ( b ) . this is well within the typical uncertainty in our imacs redshift measurements of @xmath45 km s@xmath3 @xcite , demonstrating that , along the line of sight , both galaxies are consistent with being at rest relative to the group center . secondly , both galaxies display morphological features consistent with an interaction . most prominent is an extensive stellar plume that extends @xmath45 kpc northwards from the optical nucleus of galaxy a. an extended stellar component is also visible around galaxy b , as is a bright stellar bridge between the two objects . mz10451 was observed by _ chandra _ in very faint telemetry mode for a total exposure of @xmath46 ks , with the acis - s3 ccd as aimpoint . observations were split into three separate pointings , which were analyzed using ciao v4.1 . to apply the newest calibration data and enable the suppression of background events afforded by the telemetry mode , new level 2 event files were generated . cti corrections , time - dependent gain corrections , and standard grade filtering was applied . lightcurves extracted for the s3 ccd in the 2.57 kev band revealed no evidence for background flares in any of the pointings , leaving a total effective exposure time of 78.56 ks . while the focus here is on the properties of the central group regions of mz10451 as seen by _ chandra _ , we facilitate the analysis by including our _ xmm _ data where relevant . these data were prepared following the prescription outlined in @xcite . the _ xmm _ data indicate detectable icm emission out to at least @xmath47 from the x - ray peak ( section [ sec , results ] ) , implying that any methods involving local background estimation should be avoided for the analysis of diffuse emission in the _ chandra _ s3 data . employing the `` blank - sky '' background data in the _ chandra _ calibration database for this purpose is also problematic , because the particle background component is now much more pronounced than at the time of these blank - sky observations ( see , e.g. , the discussion in @xcite ) . we have verified this by extracting blanksky - subtracted source spectra in regions that are not source - dominated , revealing prominent excess emission even at @xmath48 kev , presumably from unsubtracted particles . instead , we employed the 275-ks of ` period e ' background data taken with the acis array in the stowed position , in which only non - x - ray events are recorded . these data can thus be used to estimate the particle contribution to the total background . the latest observation to be included in these data was performed just three months prior to our mz10451 observation and so should not display serious discrepancies with respect to the actual particle level in our data . indeed , in the 9.512 kev band , the total flux ratio between source and stowed data exceeds unity by an acceptable 79% among our three pointings ( compared to almost 60% for to blanksky data ) . the stowed data are scaled up by these small factors in our analysis . in all spectral analysis , we subtracted the particle contribution using the stowed data , and modeled the remaining background components . the latter include the extragalactic x - ray background , whose spectrum can be well described by a power - law of photon index @xmath49 , and diffuse galactic emission which is strongly spatially varying but can be described by a low - temperature thermal plasma model . our _ xmm _ data were used to constrain the properties of these two components at the position of mz10451 as follows . spectra were extracted for each epic camera in bins of @xmath50 net counts within a @xmath5110@xmath52 annulus of the x - ray peak . energies @xmath53 kev were excluded , to suppress the particle - induced contribution including the instrumental fluorescence lines at @xmath541.6 kev . the results were jointly fitted with a model consisting of a power - law subject to galactic absorption plus an unabsorbed solar - abundance thermal plasma model . free parameters were thus the power - law index @xmath55 , plasma temperature @xmath56 , and the normalization of the two components . the resulting fit , with @xmath57 for 72 degrees of freedom , yielded uncertainties on the best - fit values of @xmath58 and @xmath59 kev of only @xmath60% . the pn spectrum and best - fit model are presented in figure [ fig , pnbkg ] . we include this background model in all fits to the _ chandra _ data . leaving the relative normalization of the thermal and power - law components free to vary did not impact significantly on our results , so this ratio was fixed to that derived from the _ xmm _ fit , with only the overall background normalization remaining as a free parameter . source and stowed - background spectra and associated response products were extracted separately for each of the three _ chandra _ pointings , with spectra accumulated in bins of @xmath61 net counts and jointly fitted in xspec v11.3 . whenever relevant , a similar approach was adopted for source and blanksky - background spectra extracted for each _ xmm _ camera . in addition to the x - ray data , we also acquired optical long - slit spectroscopy of each of the merging galaxies , using the imacs camera at the baade / magellan telescope in november 2009 . the f/2 camera mode was used with the 300 lines mm@xmath3 grism , giving a nominal wavelength range of 390010000 and a dispersion of 1.34 pixel@xmath3 . each galaxy was observed for 30 min using a 0.9-arcsec wide slit ( corresponding to a physical scale of @xmath62 1 kpc at the group redshift ) . general details of the flat - fielding , bias- and sky - subtraction , and wavelength calibration can be found in @xcite . the large - scale morphology of the icm in mz10451 is most easily discerned from our _ xmm _ data . based on a particle - subtracted and exposure - corrected 0.32 kev epic mosaic image , figure [ fig , xmmmosaic](a ) shows adaptively smoothed _ xmm _ contours of the central @xmath63 arcmin ( @xmath64 mpc@xmath65 ) region . the surface brightness distribution clearly exhibits a fairly symmetric morphology , consistent with the expectation for a virialized system . on the largest scales probed by _ xmm _ , there is an indication of the icm distribution being slightly elongated in the north south direction , roughly coinciding with the axis joining the two central galaxies . + based on the same ( unsmoothed ) data , figure [ fig , xmmmosaic](b ) shows a radial surface brightness profile of the diffuse emission , extracted from the x - ray peak in bins containing a signal - to - noise ratio of s / n @xmath66 . uncertainties in this plot contain a 10% systematic error associated with the subtraction of the _ xmm _ particle background , added in quadrature to all poisson uncertainties . emission is robustly detected out to at least @xmath67 arcsec ( @xmath68 kpc ) . a simple @xmath69model with @xmath70 provides a reasonable description of the profile across this radial range , with the exception of the very core which shows a slight excess of emission above the best - fit model . a thermal plasma model ( apec ) fit to the global _ xmm _ spectrum within the full region covered in figure [ fig , xmmmosaic](b ) provides an acceptable fit , @xmath71 for 111 degrees of freedom , and would suggest @xmath72 kev and @xmath73 z@xmath38 . we note that this abundance measurement is plausibly biased low due to the likely presence of temperature variations across this large radial range ( `` fe bias '' ; @xcite ) . the best - fit model would imply a 0.32 kev luminosity of @xmath74 erg s@xmath3 , with the bolometric value a factor of 1.8 higher . subtracting the particle background as estimated from stowed data and excluding identified point sources , a total of @xmath624,900 net counts ( 0.32 kev ) from diffuse x - ray emission is detected on the _ chandra _ s3 ccd , roughly a factor of four more than in the shallower _ xmm _ pn data of the same region . figure [ fig , cxomosaic ] shows a particle - subtracted , exposure - corrected , and adaptively smoothed _ chandra _ mosaic image of the central regions . afforded by the superior spatial resolution of _ chandra _ , these data clearly demonstrate that both central galaxies are offset from the peak of the diffuse x - ray emission at ( @xmath75 , @xmath76 ) = ( 02@xmath77 , @xmath78 ) , suggesting that neither galaxy is completely at rest at the center of the group gravitational potential . the x - ray morphology of the group core is quite regular , presenting little evidence that the ongoing galaxy interaction has yet had any profound impact on the central icm density distribution . also note that an emission component at the optical position of galaxy b is clearly visible . to test for radial variations in icm temperature and abundance and search for icm signatures of the ongoing merger , _ chandra _ spectra were extracted in concentric annuli centered on the x - ray peak , each containing @xmath621,000 counts from diffuse x - ray emission . these were fitted with an apec model added to the x - ray background model described in section [ sec , obs ] . deprojection was not attempted , as our _ xmm _ data show that the regions beyond the _ chandra _ s3 ccd are not free of group emission . the resulting radial profiles of @xmath79 and @xmath80 are shown in figure [ fig , profiles ] . note that the x - ray peak is located @xmath81 east of the s3 ccd center , enabling coverage towards the western ccd corners to @xmath47 in figure [ fig , profiles ] , although the angular coverage in the outermost radial bin is considerably less than 360 degrees . the temperature profile reveals the presence of slightly cooler gas in the group core , showing a peak in the radial bin that extends from @xmath821.3@xmath52 ( @xmath8390 kpc ) from the core . the central temperature drop may not be quite as pronounced as in typical cool - core groups , with the innermost sub - bin in figure [ fig , profiles ] nominally being only @xmath84% cooler than the peak temperature ( the drop is significant at the @xmath85 level ) . however , resolution effects could have an impact on this , given the higher redshift of this system compared to those of typical well - studied x - ray groups ( see , e.g. , @xcite ) . an orthogonal regression fit in @xmath79log@xmath86 space to the outermost four bins in figure [ fig , profiles](a ) yields @xmath87 for the slope outside the core , consistent with results for typical cool - core groups @xcite . we can therefore approximate the temperature profile as a piecewise log linear function , @xmath88 similarly , the abundance profile can be approximated as constant at 0.3 solar outside the core , rising linearly in log@xmath86log@xmath80 space to its central value . these parametrizations are illustrated by the dashed lines in figure [ fig , profiles ] . while the temperature profile is thus typical of cool - core groups , the abundance profile is slightly unusual . although showing the central peak of roughly solar abundance typical of such systems , the profile does not exhibit the steady radial decline seen in such groups but is instead consistent with being largely flat outside the core . note also that the derived _ abundances are everywhere larger than suggested by the fit to the global _ xmm _ spectrum within this region , supporting the suspicion that the _ xmm _ estimate could be affected by the fe bias arising from the presence of significant temperature variations . we also checked the si / fe ratio for the region encompassing the innermost two bins in figure [ fig , profiles](b ) , using a vapec model fit . although @xmath89 is poorly constrained , the ratio is found to be @xmath90 z@xmath91z@xmath92 at @xmath93 significance , indicating an important contribution to central enrichment from type ia supernovae and hence from an old stellar population . combining the _ xmm _ surface brightness profile , which is well constrained to large radii , with the parametrized _ chandra _ temperature and abundance profiles ( slightly smoothed to ensure a continuous behavior ) , profiles of deprojected gas density @xmath94 , entropy @xmath95 , cooling time , and total mass were determined . the latter was obtained via the assumption of hydrostatic equilibrium , as supported by the regular icm morphology on large scales . the results are plotted in figure [ fig , mass ] . uncertainties on each profile were obtained from 1,000 monte carlo realizations , for each of which all relevant parameters ( @xmath96 , @xmath97 , @xmath69 , @xmath98 , and surface brightness normalization ) were drawn from a gaussian distribution centered at the measured or parametrized value and with 1-@xmath99 width equal to the typical 1-@xmath99 uncertainty on the relevant parameter . at each interpolated radius , we recorded the resultant outlier - resistant biweight mean and standard deviation . the latter represent the 1-@xmath99 error bounds shown in figure [ fig , mass ] . errors on overdensity radii were evaluated from the mass profile using an analogous procedure . the entropy profile is fairly typical of cool - core groups , showing a continuous decline towards the center , and a logarithmic slope at large radii which is flatter than that expected from gravitational processes only @xcite . the cooling time profile shows a central value of @xmath100 gyr and a slope at large @xmath86 in good agreement with results for hotter systems @xcite . the derived mass profile implies a total mass of @xmath101 m@xmath38 within the radius of x - ray detection ( @xmath102 kpc ) , which is itself bracketed by the resulting values of @xmath103 kpc and @xmath104 kpc . the estimated virial radius in the adopted cosmology is @xmath105 kpc , enclosing an extrapolated mass of @xmath106 m@xmath38 . table [ tab , x_opt ] includes relevant quantities obtained from the mass profile . although mz10451 is intrinsically fairly x - ray luminous , its considerable distance ( @xmath107 mpc ) still renders it too faint to allow spectral mapping at fine spatial detail with the present _ chandra _ data . hence , to further investigate the icm properties in the group core and to identify regions of particular interest for spectral analysis , we first generated a hardness ratio map of the group core . for this , energy bands of 0.51 and 12 kev were chosen , in order to provide comparable number of counts in the ` soft ' and ` hard ' bands . we avoid the lowest energies , to reduce contamination from any patchy , soft galactic emission . mosaic images in each band were particle - subtracted and corrected for exposure variations , and results were smoothed using scales resulting from adaptively smoothing the corresponding full - band 0.52 kev image . the hard - band image was then divided by the soft - band one . for the temperatures and abundances seen in figure [ fig , profiles ] , the hardness ratio map @xmath108 in these energy bands provides a useful proxy for spatial variations in icm temperature @xmath109 along the line of sight @xcite . utilizing this , 2-d maps of projected pseudo - entropy @xmath110 and pressure @xmath111 were also constructed , based on @xmath112 and @xmath113 , where @xmath114 is the icm emission measure . to generate these maps , a 0.52 kev image , smoothed on the same scales as @xmath108 , was taken as a proxy for @xmath115 ( modulo metallicity variations ) , since at fixed metallicity @xmath116 z@xmath38 ( as seen outside the very core ) , the emissivity of a plasma varies by @xmath117% for the range of temperatures in figure [ fig , profiles ] @xcite . nevertheless , caution should be exercised when interpreting these maps , due to the degeneracy with metallicity and the presence of statistical fluctuations . the results are displayed in figure [ fig , maps ] , which shows the central @xmath118 region around the x - ray peak , corresponding to the region covered by the two innermost bins in figure [ fig , profiles](b ) . the hardness ratio map in figure [ fig , maps](a ) generally suggests fairly uniform hardness ratios on these spatial scales , but with some notable exceptions : the central regions display relatively soft emission , while a region of slightly harder emission is seen south of the group core . if interpreting these features as due to variations in icm temperature , the results support the presence of slightly cooler gas in the group core on scales of 2030 kpc , as already hinted at by figure [ fig , profiles ] . several of the detected point sources in this region also show emission which is considerably harder than that of the icm , but , interestingly , this is less obviously so for the two central galaxies . the map also suggests the presence of a curved region of slightly hotter material immediately south of galaxy b , referred to in figure [ fig , maps](a ) and in the following as the southern `` hotspot '' . the entropy map in figure [ fig , maps](b ) suggests the presence of a central region of relatively low - entropy gas , centered close to the x - ray peak . the entropy distribution generally appears fairly symmetric and reveals no prominent features around the two central galaxies . in particular , there is no clear indication of ongoing ism stripping from either galaxy , e.g. , in the form of irregularly distributed low - entropy material in their immediate vicinity . note that the southern hotspot visible in the hardness ratio map also shows up in this diagram as a region of slightly enhanced icm entropy . the pressure map , figure [ fig , maps](c ) , indicates that icm thermal pressure is generally declining outwards from the x - ray peak , as expected for hot gas in approximate hydrostatic equilibrium . however , the map also presents several potentially interesting features . one of these is a region of seemingly enhanced icm pressure _ between _ the two galaxies , as marked in the figure . no corresponding surface brightness enhancement is clearly seen ( which could suggest a local enhancement in @xmath80 at fixed projected density ) , nor in entropy or hardness ratio . we also note that the x - ray peak seems to be surrounded by two small low - pressure regions , with no obvious counterparts in the surface brightness distribution ( cf.figure [ fig , cxomosaic ] ) , but the available statistics do not allow a detailed investigation of their robustness or possible origin . we can , however , test the robustness of the hotspot seen south of galaxy b. for this , a spectrum was extracted in a rectangular region covering this feature and fitted with a thermal plasma model . figure [ fig , specregions ] outlines the spectral extraction regions discussed in the following . within the region labelled `` h '' in the figure , the derived temperature , @xmath119 kev , is indeed higher than that obtained for a comparable control region at similar distance north of the x - ray peak ( region `` c '' in figure [ fig , specregions ] ) , @xmath120 kev . formally , these results differ at 1.5-@xmath99 significance , with the southern region being @xmath84% hotter . we note that although the position of the temperature peak in figure [ fig , profiles ] is consistent with that of this region , results for the radial profile do not change significantly if excluding this region from the second radial bin in the profile ; the presence of a peak in the temperature profile at this radius is therefore not simply due to the hotspot . we also note that the derived metallicity within this region , @xmath121 z@xmath38 , is fully consistent with the radial average at this distance from the x - ray peak , @xmath122 z@xmath38 . this implies that the presence of this feature is not simply due to strong local variations in icm abundance . another interesting question is how the icm metallicity in the region covered by the stellar plume of galaxy a compares to that of its immediate surroundings . this plume might represent tidally stripped material that could evolve into an intracluster light component . these stars would be able to chemically pollute the surrounding icm very efficiently , as their ejecta would mix directly with the icm without having to overcome the confining gravitational potential and gas pressure of their former host galaxy . to test this possibility , a spectrum was extracted in a wedge extending @xmath123 northwards from the x - ray peak ( region `` p '' in figure [ fig , specregions ] ) , with the minimum size of the region dictated by the need to get at least 500 counts as required for a rough measurement . corresponding wedges extending to the east and west of the peak were employed as control regions . all regions had their apex centered on the x - ray peak rather than on galaxy a itself , such that the metal - rich group core ( cf.figure [ fig , profiles ] ) would be equally represented in all wedges and thus not bias the result for the northern wedge . for the n wedge , we find @xmath124 z@xmath38 , compared to @xmath125 z@xmath38 for the combined e and w wedges . hence , gas coinciding with the stellar plume does show significantly higher abundance than other regions at similar distance from the group core . this supports a scenario in which the central rise in the radial abundance profile has a significant contribution from highly enriched gas coinciding with this plume . turning now to the properties of the interacting galaxies , an important question is whether the galaxies have retained any hot gas , and whether the ongoing interaction has triggered significant starburst or active galactic nucleus ( agn ) activity in either . galaxy b is particularly interesting in this regard , because our _ spitzer _ data reveal @xmath12 m emission from this galaxy @xcite . to explore these issues , we present magellan optical spectra of both galaxies in figure [ fig , opt_spec ] ( obtained with a slit width covering the central @xmath126 kpc of each galaxy ) , along with our near - uv and @xmath12 m data of the pair in figure [ fig , galpair ] . the optical spectra appear remarkably similar , and neither shows clear evidence for bright emission lines typically associated with star formation or nuclear activity . this is in stark contrast to results showing that practically all ( fully formed ) bcgs close to the x - ray center in cool - core clusters display optical line emission @xcite . also note that the nucleus of galaxy b is actually the brighter at most optical wavelengths , whereas that of galaxy a dominates towards the near - uv as also confirmed by our _ galex _ data . given the lack of bright optical emission lines in galaxy b , its @xmath12 m emission may be attributed to either obscured agn activity or dusty star formation . in contrast , the optically brighter galaxy a is not detected above @xmath127 significance at @xmath12 m . nevertheless , extended emission from both galaxies is detected in the _ galex _ nuv band at @xmath128 significance . this may suggest the presence of some recent or ongoing low level star formation , although the nuv@xmath129 colors of 5.5 ( galaxy a ) and 6.1 ( b ) would render both galaxies `` quiescent '' by some definitions ( e.g.@xcite ) . in addition , based on @xmath130 magnitudes from the 2df survey , both galaxies have comparable optical colors of @xmath130@xmath131 ( a ) and @xmath132 ( b ) , consistent with typical passive early - types . neither galaxy is detected in the far - uv , implying fuv nuv@xmath133 in either case , nor at @xmath134 m in our _ spitzer _ data . they are also not detected at 1.4-ghz in nvss radio data , which are 99% complete down to @xmath135 mjy @xcite , implying 1.4-ghz luminosities below @xmath136 w hz@xmath3 . finally , to estimate their stellar masses , we note that 2mass isophotal photometry is available for galaxy b only , implying @xmath137 l@xmath38 for this galaxy . if assuming a @xmath138-band stellar mass - to - light ratio @xmath139 m@xmath140l@xmath38 , then @xmath141 m@xmath140l@xmath38 for galaxy b. assuming a similar relationship for galaxy a ( see also @xcite ) yields the approximate stellar masses listed in table [ tab , gals ] which summarizes some relevant properties of the two galaxies . note that the stellar mass estimate for galaxy a includes the diffuse stellar plume . lcc & galaxy a & galaxy b + ra ( j2000 ) & 02@xmath142 & 02@xmath143 + dec ( j2000 ) & @xmath144 & @xmath145 + @xmath146 ( l@xmath38 ) & @xmath147 & @xmath148 + @xmath149 ( l@xmath38 ) & @xmath150 & @xmath151 + @xmath152 & 20.62 & 21.97 + @xmath153 ( erg s@xmath3 ) & @xmath154 & @xmath154 + @xmath155 ( w hz@xmath3 ) & @xmath156&@xmath156 + @xmath157 ( m@xmath38 ) & @xmath158 & @xmath159 + based on our wavelet - based 0.45 kev point source search , galaxy a is not detected in the _ chandra _ data , possibly due to its proximity to the central x - ray peak , whereas galaxy b is clearly detected . this is true also if only considering a soft band ( 0.32 kev ) source search , whereas none of the galaxies is detected at higher energies such as 28 kev . galaxy b is not robustly detected in 28 kev _ xmm _ data either . all this immediately argues against significant moderately obscured agn activity in either galaxy , and potentially also against the presence of a significant hot gas component in galaxy a. for galaxy a , any x - ray surface brightness analysis is severely hampered by its proximity to the peak of the icm emission @xmath160 to the north . if employing a 5@xmath161 semi - annulus to the south as representative of the local background around the source , it is formally detected at @xmath162 above this background in 0.32 kev . this is below the @xmath127 significance threshold adopted for the adaptive smoothing in figure [ fig , cxomosaic ] , explaining its apparent absence in that figure . it is not detected in 13 kev however , implying a background - subtracted ( 13 kev)/ ( 0.51 kev ) hardness ratio of @xmath163 at 90% confidence . this in turn implies @xmath164 for a power - law spectrum subject to galactic absorption , or @xmath165 kev for a thermal plasma with @xmath166 z@xmath38 . this suggests that any emission from this source has properties inconsistent with those of a typical agn spectrum , but leaves open the possibility of some contribution from hot gas . as further confirmation of the lack of substantial nuclear x - ray activity , we note that the 0.58 kev count rate within @xmath167 implies @xmath168 erg s@xmath3 in this band for any intrinsically unobscured @xmath169 component . furthermore , since @xmath167 still corresponds to a physical scale of @xmath170 kpc , this upper limit contains a contribution from non - nuclear emission . we also note that we do not detect any point sources in the stellar plume of this galaxy , down to a limiting 0.52 kev flux of @xmath171 erg @xmath10 s@xmath3 ( for a @xmath172 spectrum ) . this implies that there are no clusters of x - ray binaries in the plume with a combined luminosity exceeding @xmath173 erg s@xmath3 , thus also ruling out the presence of any ultra - luminous x - ray sources in this region . in contrast to the case of galaxy a , there is a clear indication that the soft emission associated with the optically fainter galaxy b is spatially extended in the _ chandra _ data . figure [ fig , galb ] shows a 0.32 kev surface brightness profile extracted from the optical galaxy center in bins of s / n @xmath174 . emission is detected out to @xmath175 at @xmath127 above the local background , as evaluated in a surrounding @xmath176@xmath177 annulus . both source and background regions excluded position angles of @xmath178 around the north , to suppress potential contamination from galaxy a and from local variations in icm emission . the inferred profile is clearly much broader than that of the @xmath179 kev _ chandra _ point spread function ( psf ) at this detector position as estimated using the _ mkpsf _ tool in ciao . a kolmogorov smirnov test yields a probability of only @xmath180 that the two data sets have been drawn from the same distribution , and the source profile is well described by a @xmath69model with @xmath181 as is typical for extended emission from early - type galaxies ( e.g. , @xcite ) . at 95% confidence , a simple absorbed power - law fit to the spectrum of galaxy b within @xmath175 would require @xmath182 ( for spectral bins of 15 net counts , assuming standard @xmath183 statistics ) or @xmath184 ( assuming cash statistics with bins of 5 net counts ) . in addition , the fit shows obvious positive residuals at @xmath185 kev , strongly suggestive of an fe peak . these results are inconsistent with the expectation for emission dominated by an agn or low - mass x - ray binaries ( which both tend to show @xmath186 ) , but suggest an important contribution from thermal gas . an apec model with galactic absorption returns a best - fit @xmath187 kev for @xmath80 fixed at 0.5 z@xmath38 , with @xmath188 erg s@xmath3 ( 0.32 kev ) . for these parameters , the observed flux would imply a total hot gas mass of @xmath189 m@xmath38 within @xmath175 ( @xmath190 kpc ) . within @xmath167 of the optical center , the ( 28 kev)/ ( 0.52 kev ) particle - subtracted hardness ratio of @xmath191 also suggests rather soft emission . this result is difficult to reconcile with the presence of an intrinsically bright but moderately obscured source , since any intrinsically obscured power - law source with @xmath192 should display a ratio of @xmath193 . with galactic absorption only , the observed ratio would instead suggest @xmath194 ( @xmath195 at 1-@xmath99 ) , inconsistent with a typical agn spectrum . the total 0.58 kev count rate within @xmath196 happens to be identical to that of galaxy a , again implying @xmath168 erg s@xmath3 for any intrinsically unobscured @xmath172 agn . to escape _ chandra _ detection in the 28 kev band , any agn intrinsically brighter than @xmath197 erg s@xmath3 with a @xmath172 spectrum must be subject to an intrinsic absorbing column in excess of @xmath198 @xmath10 , and the source would have to be compton - thick ( @xmath199 @xmath10 ) for all intrinsic luminosities @xmath200 erg@xmath3 . in summary , galaxy a which remains undetected at @xmath12 m shows little evidence for a substantial agn component , unless sufficiently obscured to evade detection even at energies up to 8 kev . even for the ir - detected galaxy b , there is also no evidence for prominent nuclear x - ray activity below this energy . the lack of significant hard x - ray emission within the _ chandra _ energy range in either object is also supported by the hardness ratio map , in which both galaxies appear softer than typical point sources in the field ( some of which are likely associated with background agn ) . regarding any hot ism material , tentative evidence for the presence of hot gas is found in galaxy a , but we caution against strong conclusions in this case , given that this galaxy is superposed on very bright icm emission . however , galaxy b provides clear evidence for extended emission probably associated with hot `` halo '' gas . for its @xmath130@xmath129 color of 2.1 derived above , even extreme estimates of the discrete source contribution to the total @xmath26 in early - types of the resulting @xmath201 would suggest that @xmath202% of its x - ray output is due to hot gas @xcite . furthermore , its @xmath26 within the @xmath203 radius of detection would place it above , but within the scatter of , the relation between @xmath201 and @xmath26 for local early - types @xcite . the same applies when comparing it to the expectation from the relation between @xmath204 and thermal @xmath26 for early - types in groups as derived by @xcite . these results are all consistent with the idea that galaxy b contains significant amounts of hot diffuse gas . the icm x - ray luminosity and emission - weighted mean temperature derived from the _ xmm _ data , as listed in table [ tab , x_opt ] , would place mz10451 very close to the @xmath26@xmath79 relation for x - ray bright groups of @xcite . combined with the regular x - ray morphology on both small and large scales , and the clear evidence for a cool core and a central metallicity excess seen in figures [ fig , profiles ] and [ fig , maps ] , these features suggest a system with icm properties very typical of those of relaxed x - ray bright groups and clusters . before addressing the impact of the central galaxy merger on the icm , as well as the nature of the merger itself , it is useful to first assess the current stage of the interaction . it is not immediately clear from the present data whether the two galaxies have already experienced a first close passage . the pronounced tidal stream protruding northwards from galaxy a might suggest so , but dedicated @xmath205-body simulations of binary mergers ( which we hope to employ in future work ) would be needed to firmly establish under which orbital conditions this feature could arise . one argument _ against _ this possibility , however , is the observation of an extended thermal component associated with galaxy b. it seems unlikely that a galactic hot gas halo would survive a direct passage of one galaxy through the other , especially since galaxy b would then have approached from the north and so also have passed through the dense icm core . from the available information we can provide a rough estimate of the time - scale for nuclear coalescence . to this end , we apply the calibration of @xcite based on the millennium simulation , @xmath206 where @xmath157 is the stellar mass of the merger product and @xmath207 the current projected distance between the progenitors . using @xmath208 m@xmath38 from the numbers in table [ tab , gals ] along with @xmath209 kpc , one obtains @xmath210 myr . hence , the galaxies are likely still at least half a gyr from coalescence , a number that is probably uncertain by at least a factor of two . table [ tab , time ] summarizes , in ascending order , this and other timescales discussed in the following . rll timescale & description & see + @xmath211 myr & timescale for dispersal of icm pressure enhancement between galaxy a and b & section [ sec , impact ] , figure [ fig , maps](c ) + @xmath212 myr & age of southern hotspot if representing a weak shock front propagating from icm core & section [ sec , merger ] , figure [ fig , maps](a ) + @xmath213 gyr & central icm cooling time & section [ sec , impact ] , figure [ fig , mass](c ) + @xmath214 gyr & timescale for generation of central icm metal excess by stellar plume of galaxy a & section [ sec , impact ] + @xmath215 gyr & timescale for nuclear coalescence of galaxy a and b & equation ( [ eq , tmerge ] ) + @xmath216 gyr & dynamical ( free - fall ) timescale at @xmath32 & section [ sec , merger ] + @xmath217 gyr & time since last major group group merger in mz10451 from x - ray isophotes & section [ sec , merger ] ; @xcite + @xmath218 gyr & icm sound crossing time within @xmath32 & section [ sec , merger ] + @xmath219 gyr & timescale for mz10451 to evolve into a `` fossil group '' & equation ( [ eq , dynfric ] ) + the regular x - ray morphology of mz10451 , even in the central regions , suggests that the ongoing galaxy interaction has not had any major impact on the icm so far . in particular , although the merger has significantly disrupted the stellar content near the center of mz10451 , the system seems to have retained a cool core at this stage . the possible destruction of cool cores , or the prevention of them forming in the first place , is still poorly understood and widely debated in the literature ( e.g. , @xcite ) . our results suggest that central galaxy galaxy mergers do not necessarily completely destroy cool cores through icm heating or mixing , or that any such activity occurs closer to the final merger stages ( when perhaps quasar - like activity is triggered in the case of gas-_rich _ major mergers ; cf . @xcite ) . in partial support of the former possibility , we note that dynamical heating by galaxy motions alone generally seem insufficient to significantly disrupt cooling in cluster cores @xcite . the situation in mz10451 is somewhat different from that in typical x - ray clusters where a fully formed bcg is already residing at the cluster center and potentially re - heating its surroundings via intermittent agn activity . nevertheless , the offset between galaxy a and the x - ray peak is still small , @xmath220 kpc , and the central cooling time of @xmath213 gyr is comparable to that of other cool - core systems at the relevant radii @xcite . this further supports the idea that any icm mixing must have been modest in the very core of mz10451 . this is also borne out by the entropy distribution ; while enhanced central entropy compared to the case of typical cool - core systems could suggest a recent heating or mixing episode , we find the central entropy to be quite low , and close to the @xmath221 kev cm@xmath65 below which strong h@xmath75 and radio emission is typically seen in central bcgs @xcite , presumably as a result of ongoing star formation fueled by radiative cooling of icm material . despite the apparent absence of strong icm mixing , some features of the icm thermodynamic properties may be linked to the ongoing interaction . figure [ fig , maps](c ) indicates the presence of a region of enhanced icm pressure between the two galaxies , with no obvious counterpart in the surface brightness or entropy maps . in the absence of external influences , this feature should disappear roughly on its sound crossing timescale , which is @xmath211 myr for its estimated spatial extent of @xmath223 kpc . it is therefore likely to be associated with a very recent or ongoing event , plausibly reflecting ongoing adiabatic compression and heating of the icm between the galaxies , induced by their relative motion . it remains unclear , however , whether the tentative presence of an extended region of hot , high - entropy gas immediately south of galaxy b is related to the interaction . the hardness ratio and entropy maps suggest that the extent of the structure is several tens of kpc even at its narrowest range . thus , an explanation invoking a shock front generated by galaxy b moving supersonically southwards seems excluded . alternatively , this gas could represent high - entropy material ejected or stripped from either galaxy as a result of the interaction , but again , the sheer size of this feature makes this unlikely . there are insufficient counts in this region to test for departures from solar abundance ratios , thus precluding any direct tests for a possible origin in starburst outflows . however , such an explanation would be at odds with the lack of clear post - starburst features in the optical spectrum of either galaxy . it is possible that the feature is due to `` sloshing '' of the core gas owing to the gravitational perturbation to the group potential caused by the ongoing merger , although the absence of a similar feature in the surface brightness distribution may argue against this @xcite . another possibility , discussed in more detail in the next section , is that it is related to a larger - scale interaction between two sub - groups . from the viewpoint of the icm abundance distribution , figure [ fig , profiles ] shows that any merger induced icm enrichment is so far confined to the innermost @xmath224 kpc and so has had no large - scale impact on the icm metal content . nevertheless , the central abundance peak seen in this and many other systems could be partly related to processes associated with bcg formation . the observation of enhanced icm abundances in the region covered by the stellar plume of galaxy a may signify the presence of stripped , highly enriched ism from this galaxy . this would suggest that central metal excesses could generally have a contribution from enriched gas shed by central bcg progenitors . in fact , the icm mixing expected during such a merger may help to explain the considerable _ extent _ of the central abundance peak in many systems , thus alleviating the need for additional subsequent processes , such as agn driven turbulence @xcite , to disperse metals well beyond the optical extent of the central bcg . an alternative explanation would associate the central abundance peak in mz10451 with _ in situ _ enrichment by stars tidally stripped during the interaction , some of which are likely to evolve into an intracluster light component . to assess the feasibility of this scenario , we first reiterate that the subsolar si / fe ratio seen in the central region implies an important contribution from sn ia and hence from old stars , in line with results for fully formed bcgs in x - ray bright groups @xcite . this also indicates that central icm enrichment from any young stellar population in the plume ( or within the possibly star - forming galaxy b ) must be modest , in line with the apparent absence of the plume in our uv and @xmath12 m images in figure [ fig , galpair ] . at the resolution of figure [ fig , profiles ] , the elevated icm abundance in the group core implies an _ excess _ fe mass in the core of @xmath225 m@xmath38 . aperture photometry based on our magellan @xmath129-band image further suggests that @xmath226 of the stellar light of galaxy a is located in this plume , indicating a plume stellar mass of @xmath227 m@xmath38 for @xmath228 m@xmath140l@xmath38 as assumed in section [ sec , pair ] . given these values , sn ia in the plume could have provided the central fe excess on a timescale of just @xmath215 gyr , estimated using the approach of @xcite and excluding any contribution from sn ii or stellar winds . thus , it seems entirely feasible for intracluster light generated by the merger to have produced the central fe excess . forthcoming _ hubble space telescope _ ( _ hst _ ) imaging of mz10451 will be used to provide more reliable mass and age estimates of the stripped stellar component of galaxy a and further test the viability of this picture . the relatively undisturbed appearance of the icm suggests that the central bcg merger does not result as a consequence of a recent larger - scale merger between separate subgroups . idealized hydrodynamical simulations of major cluster cluster mergers with a range of progenitor mass ratios and impact parameters suggest that the timescale for a merged system to appear relaxed , as judged visually from the morphology of its x - ray isophotes , is @xmath217 gyr from the time at which the gas core of the secondary merges with that of the primary @xcite . for mz10451 specifically , we note that the icm appears regular on the scales of figure [ fig , xmmmosaic ] , i.e.out to at least @xmath229 kpc . the icm should relax roughly on a sound crossing time @xmath230 . using equation ( [ eq , tprof ] ) to evaluate the local sound speed at all @xmath86 gives @xmath231 gyr within @xmath232 kpc , and @xmath233 gyr within @xmath234 kpc , in broad agreement with the general expectation from the above simulations . this confirms that any major group group merger in mz10451 can not have taken place very recently . however , it may still be premature to entirely dismiss the possibility of such a merger , partly because the observed entropy enhancement to the south of the group core is not easily explained by a galaxy galaxy interaction alone , and partly because the icm may have relaxed more rapidly than the galaxies . we note that the entropy feature is not seen in the surface brightness distribution of figure [ fig , cxomosaic ] , is not particularly sharply defined in the entropy and temperature maps of figure [ fig , maps ] , and has higher temperature and entropy than its surroundings . it is therefore unlikely to represent a cold front @xcite resulting from a subgroup penetrating the group core , or a remnant cool core of such a subgroup . nevertheless , it is still conceivable that the feature represents a weak , merger - generated shock front , possibly viewed at some oblique angle . if so , a lower limit to its age of @xmath235 myr can be estimated from its current projected distance to the x - ray core of @xmath236 kpc , assuming propagation from the core at a velocity comparable to the local sound speed . one might have expected a similar feature to the north of the group core , but projection effects could be rendering this undetectable . in addition , simulations of cluster cluster mergers @xcite suggest that such central entropy asymmetries , including the curved appearance of the region in figure [ fig , maps ] , _ can _ arise as a consequence of shock heating in cluster mergers when the progenitor mass ratio differs substantially from unity ( @xmath2373:1 ) . we note that dissipation of turbulent energy injected by such a shock could easily account for the observed entropy enhancement . the mean icm density at the approximate radial distance of this feature is @xmath238 @xmath239 ; for a cylindrical volume of length @xmath240 and base diameter @xmath241 ( assuming a depth along the line of sight equal to the `` width '' of the feature ) , the total gas mass is then @xmath189 m@xmath38 within this region . the energy required to raise @xmath79 of this gas by @xmath242 kev , as suggested by our spectral analysis , is @xmath243 erg . this is three orders of magnitude below what can be provided by shocks generated by a major group group merger producing a system of the relevant virial mass @xcite , so energetically it seems entirely feasible to produce this feature through shocks associated with an infalling subgroup . hence , the observed north south entropy asymmetry could conceivably result from the aftermath of a group group merger , with galaxy a and b representing the central galaxies of the two progenitor groups . it is , in fact , possible that galaxy a was initially at rest at the x - ray peak of the more massive subgroup but has now been slightly displaced as a result of the final central galaxy galaxy interaction . there are arguments both for and against this interpretation . on the one hand , the estimated time since the last major merger in mz10451 of @xmath244 gyr , as discussed above , is larger than both the dynamical ( free - fall ) timescale at @xmath32 of @xmath216 gyr and the current estimated timescale for nuclear coalescence between galaxy a and b ( cf . table [ tab , time ] ) . hence , if each galaxy represented the central galaxy of a progenitor system , it is perhaps surprising that they have not yet themselves merged . on the other hand , we note that the galaxy velocity distribution in mz10451 does appear bimodal , with the velocity histogram of the group members , shown in figure [ fig , galdistr](a ) , revealing a smaller , receding peak relative to the system mean ( containing 12 of the 60 confirmed group members ; @xcite ) . as also illustrated in figure [ fig , galdistr](a ) , the spatial distribution of galaxies in the group provides no clear indication that infall of a subgroup , let alone a group group major merger , has recently taken place . to test this possibility in more detail and search for evidence of kinematic substructure in the system , we performed a dressler shectman test @xcite . for each galaxy @xmath245 with radial velocity @xmath246 , we recorded the local kinematic deviation @xmath247 from the global mean velocity @xmath248 and velocity dispersion @xmath249 , @xmath250,\ ] ] using @xmath251 and @xmath252 derived including its 10 nearest neighbors . we also recorded the cumulative deviation @xmath253 , @xmath254 calibrating this result against 10,000 monte carlo simulations in which all galaxy velocities @xmath246 were randomly shuffled . the latter step provides an estimate of the probability @xmath255 of obtaining the observed galaxy position velocity configuration in the presence of purely random substructure . we find @xmath256% , regardless of whether all 60 members or only the 19 covered by the detectable x - ray emission are included in the calculation . the results for @xmath257 are illustrated in figure [ fig , galdistr](b ) and suggest the presence of some non - random kinematic substructure in mz10451 , both globally and in the central x - ray bright regions . thus , the observed galaxy kinematics seem consistent with the idea that infall of a smaller subgroup ( possibly hosting galaxy b as the central galaxy ) has taken place . we also note that the estimated timescales for several of the features in figures [ fig , profiles ] and [ fig , maps ] to arise or survive ( the cool core , central metal excess , southern hotspot , and pressure enhancement between galaxy a and b ) are all consistent with being lower than the likely time since such a merger ( table [ tab , time ] ) , so their presence seems compatible with this possibility . however , we stress that this interpretation must be regarded as tentative , is not without its problems , and that more robust tests of the group group merger scenario would require detailed spectral mapping of the group core and hence deeper x - ray observations . in the @xmath129-band , the two central galaxies show a luminosity ratio of 2:1 , with the brighter galaxy a having a luminosity comparable to an @xmath258 galaxy @xcite . they currently represent the first and fifth - ranked group members within the derived virial radius of @xmath259 mpc , although brighter galaxies with concordant redshifts are present within @xmath260 from the x - ray peak . however , the combined @xmath129-band luminosity of the central pair , @xmath261 l@xmath38 , is @xmath262% larger than that of any of those galaxies within this radius . hence , the merger will comfortably produce the brightest galaxy within the group , thus confirming it as a bcg major merger . recent _ chandra _ observations show that most luminous early - type galaxies maintain thermal x - ray halos even near the centers of groups and rich clusters @xcite . for example , excluding central bcgs , @xcite found that @xmath46% of early - type galaxies in groups with @xmath263 l@xmath38 retain a hot gas halo . both our galaxies have @xmath204 comparable to this limit ( section [ sec , pair ] ) . the presence of extended thermal emission associated with galaxy b indicates that it , too , possesses a hot halo , suggesting that such halos may be preserved at least at this merger stage . in the case of galaxy a , no such halo is unambiguously detected , so it remains a possibility that any extended halo of this galaxy has been largely disrupted , since this galaxy also appears particularly disturbed in the optical . we should note , however , that we do not have the s / n or spatial resolution to directly establish whether either galaxy shows current evidence for hot gas stripping , although the pseudo - entropy map ( figure [ fig , maps ] ) suggests that any such activity must be modest . it is commonly assumed , as also suggested by numerical simulations , that the nature of low - redshift major mergers leading to bright , early - type bcgs is largely dissipationless , involving progenitors containing little or no gas ( e.g. , @xcite ) . based on observed stellar rotational velocities in nearby bcgs , some likely counterexamples do exist however , although these seem rather rare @xcite . it is therefore of interest to establish to what extent either of the merging galaxies in mz10451 is experiencing active star formation fueled by a reservoir of _ cold _ gas . in the case of galaxy a , the observed nuv flux would suggest weak star formation at a level of @xmath264 m@xmath38 yr@xmath3 , but no ir emission is accompanying the uv output down to our approximate detection limit of 0.1 m@xmath38 yr@xmath3 @xcite . in addition , its fuv nuv color of @xmath265 is consistent with that seen for the majority of passively evolving ellipticals @xcite , suggesting that the uv light may be dominated by that of an evolved stellar population . while this latter point remains true also for galaxy b , this case is nevertheless more ambiguous . the galaxy is detected at 24@xmath266 m , but there is no clear evidence for optical emission lines or nuclear x - ray activity . any agn responsible for the 24@xmath266 m emission must thus be subject to strong intrinsic obscuration , with an absorbing column exceeding @xmath267 @xmath10 for typical agn x - ray luminosities . examples of such highly obscured agn do occur at low redshift ( see , e.g. , @xcite ) , including , as is the case for galaxy b , galaxies that do not show any seyfert signatures in the optical band . alternatively , the infrared emission may instead be powered by dusty star formation , implying that the galaxy also retains some cold gas . if so , the observed 24@xmath266 m flux would suggest low - level star formation at a rate of @xmath264 m@xmath38 yr@xmath3 , computed using the prescription in @xcite which takes into account the contribution from cold dust heated by an evolved stellar population . if attributed to young stars , the _ galex _ nuv flux would instead suggest a rate of @xmath268 m@xmath38 yr@xmath3 based on the @xcite relation and before correction for dust attenuation . this is consistent with the idea that some obscured star formation activity is taking place , supporting the notion of a cold gas reservoir in galaxy b. unfortunately , the absence of a robust _ spitzer _ detection of either galaxy at 70@xmath266 m implies that we can not use the 24/70@xmath266 m flux ratio to distinguish between star formation and agn activity . furthermore , the available nvss radio data of the galaxy pair are not deep enough to provide useful constraints in this regard . the radio luminosity limits are well below a commonly adopted dividing line between starburst galaxies and radio - loud agn of @xmath269 w hz@xmath3 ( e.g. , @xcite ) , but while the limit on the @xmath270 parameter for galaxy b , @xmath271 , is consistent with that of star - forming galaxies ( e.g. , @xcite ) , it does not rule out a contribution from agn activity to the 24@xmath266 m emission . deep hi or co observations would thus be needed to decisively test for the presence of a cold gas reservoir in galaxy b. while such data will eventually be provided by the upcoming wallaby survey to be undertaken by the australian ska pathfinder array , we simply note here that the tentative evidence for obscured star formation within galaxy b points to an early - type major merger which does involve some gas , with the implication that at least some low - redshift bcg major mergers must be dissipational to some degree . coupled with the absence of pronounced nuclear activity in either galaxy , this result is consistent with a picture in which the early stages of interactions between early - type galaxies first trigger residual star formation , followed by efficient central inflow of gas leading to strong agn activity at the final stages ( e.g. , @xcite ) . if identifying the peak of diffuse x - ray emission with the center of the group gravitational potential , our results also suggest that bcgs may form very close to the latter . this is consistent with a picture in which central bcgs have been built up through a series of mergers near cluster cores , in line with the galactic cannibalism scenario for bcg growth discussed by @xcite . as mentioned , it is possible that galaxy a was initially at rest at the position of the x - ray peak but has now been slightly displaced as a result of the ongoing interaction . furthermore , our radial velocity measurements and the fact that the merging pair and the x - ray peak are all aligned , indicate that the motion of the galaxies occurs largely in radial orbits along a north - south axis , coinciding with the direction defined by the stellar plume of galaxy a. at the same time , the large - scale x - ray morphology seen in figure [ fig , xmmmosaic ] implies an overall icm elongation along a similar axis . if the elongation of the icm betrays the orientation of the filaments feeding the cluster , then the merging galaxies and their `` parent '' filaments line up . this is consistent with expectations from simulations , which suggest that the mergers forming central bcgs occur preferentially along radial orbits aligned with the filaments feeding the host cluster , and thus with the major axis of the cluster itself ( @xcite and references therein ) . as a final point , it is interesting to also consider the likely properties of the merger product , in terms of the nature of the merged central galaxy and of the group as a whole . as mentioned , the estimated @xmath138-band luminosity of the merged galaxy would be @xmath272 l@xmath38 . using @xmath39 from table [ tab , x_opt ] , we can compare this to the expectation based on the observed low - redshift correlations between bcg @xmath138-band luminosity and @xmath39 of the host cluster derived by @xcite and @xcite . either of these relations would suggest @xmath273 l@xmath38 for a bcg in a group of this mass , in excellent agreement with the above estimate . while some of the stellar mass of the bcg progenitors will be tidally stripped in the merging process and thus will not contribute directly to @xmath274 @xcite , it seems likely that mz10451 in the post - merger phase would thus have properties in good agreement with the observed @xmath275@xmath39 relations , particularly in light of the very considerable scatter around these relations at low values of @xmath39 . the tidally stripped stars seen around both merging galaxies may become unbound during the interaction and would then contribute to the emergence of an intracluster light component . this possibility will be investigated in more detail in a future paper based on our forthcoming _ hst _ imaging of this system . however , we note for now that our existing imaging of mz10451 seems consistent with the notion that an intracluster light component can form in systems which do not already have a central dominant early - type galaxy . such a scenario draws support from cosmological simulations which suggest that major mergers associated with bcg formation play a prominent , and possibly dominant , role in the generation of intracluster light @xcite . it is also interesting to note that mz10451 would not become a `` fossil group '' as a consequence of the merger , if adopting the definition of @xcite . although the x - ray luminosity criterion , @xmath276 erg s@xmath3 , is already safely met , the combined magnitude of the merger product of @xmath277 would be only @xmath278 mag brighter than that of the subsequently second - brightest galaxy within @xmath279 kpc , thus falling short of the @xmath280 gap between the first- and second - ranked galaxy required to meet the above definition of a fossil system . this remains true even if just considering the region within @xmath281 kpc , for which @xmath282 would be @xmath283 mag . taking this argument even further , we note that any subsequent merger between the forming central galaxy and the resulting brightest satellite galaxy within a projected distance of @xmath284 ( with @xmath285 m@xmath38 ) _ would _ create an @xmath129-band magnitude gap of @xmath286 ( @xmath287 ) . this satellite currently has a radial velocity @xmath288 km s@xmath3 relative to the group mean and so is likely to be in a low - eccentricity orbit . under these circumstances , dynamical friction acting on this galaxy could accomplish such a merger on a time - scale of @xmath289 where @xmath290 and @xmath291 is the current orbital radius and circular velocity of the satellite , respectively , and @xmath292 the characteristic circular velocity of its dark matter halo @xcite . a lower limit to @xmath290 is the projected distance of the satellite from the x - ray peak of @xmath293 kpc . taking @xmath294 and assuming an @xmath129-band total mass - to - light ratio of 10 for the satellite ( a spiral ) , along with the assumption of an isothermal sphere mass distribution for its halo , we have @xmath295 km s@xmath3 . with these numbers , equation ( [ eq , dynfric ] ) would predict a time - scale @xmath296 gyr , with equality corresponding to @xmath297 . thus , even following the ongoing central merger , mz10451 would be resilient to the formation of a fossil - like configuration by means of dynamical friction for many hubble times . this is consistent with the notion that present - day fossil systems are unlikely to have formed recently ( e.g. , @xcite ) . the @xmath298 low - mass cluster mz10451 is , to our knowledge , a unique system . it is x - ray bright , harbors a central bcg currently being formed from the major merger of two optically luminous early - type galaxies , and resides at a distance where the nature of the bcg merger and its impact on the surrounding icm can be studied in unprecedented detail . we have combined _ chandra _ observations of the icm in the cluster core with _ xmm _ data of the large - scale icm properties , and with optical , infrared , and uv data of the merging galaxies , to explore the impact of the interaction on the icm and on the interacting galaxies themselves . the data indicate detectable icm emission out to at least @xmath299 , centered close to the two interacting galaxies . the icm surface brightness in mz10451 is symmetrically distributed on both small and large scales , consistent with the expectation for a virialized system . hence , the ongoing central galaxy galaxy interaction has not significantly disturbed the icm density distribution even in the group core . on the largest scales , the icm distribution is elongated in the north - south direction , however . the icm exhibits a cool core and a central metal excess ( here extending to @xmath300 kpc ) , as typically seen in undisturbed x - ray bright systems with a central bcg . this implies either that cool cores and central abundance peaks are not necessarily destroyed by central galaxy galaxy mergers in clusters , or that any significant central icm heating and mixing takes place at post - coalescence stages . the brighter of the merging galaxies presents a pronounced plume of tidally stripped stars . the icm metallicity in the region covered by this feature is significantly enhanced . this suggests efficient local enrichment by intracluster stars ejected by the interaction , or merger - induced stripping of highly enriched gas from either galaxy . such processes may thus have contributed to building up the pronounced central abundance peak seen in this and other systems . some of the tidally stripped stars may represent the emergence of an intracluster light component , suggesting that such a component can form in systems that do not already host a central dominant galaxy . both the merging galaxies are spatially offset from the icm x - ray peak , but only by @xmath220 and @xmath221 kpc , and their radial velocities are both consistent with the group mean within the errors . this demonstrates that central bcg formation can occur close to the center of the cluster potential , consistent with the notion that central bcgs are built up through a series of mergers in cluster cores . the merger axis extends toward the x - ray peak and roughly coincides with the axis defined by the large - scale elongation of the icm . combined with the above , this provides support for a picture in which bcg mergers are caused by galaxies falling along radial orbits aligned with the accreting filaments that feed cluster growth . a rough estimate based on their current separation suggest that the interacting galaxies are still @xmath301 gyr from nuclear coalescence . despite being involved in a galaxy galaxy interaction in the core of an x - ray bright system , at least one of the merging galaxies shows clear evidence of having retained an extended hot gas halo at this stage . the inferred infrared , uv , and x - ray properties of the galaxy pair are consistent with the presence of low - level obscured star formation in at least one of them . this would be consistent with a picture in which early - type galaxy interactions initially trigger residual star formation , followed by an agn - dominated phase closer to coalescence . the tentative evidence for ongoing star formation in one of the merging galaxies suggests that low - redshift bcg major mergers need not be completely dissipationless . however , while the x - ray data reveal no evidence for strong moderately obscured agn activity in either galaxy , the possibility of an agn hidden behind a compton - thick screen can not be ruled out as an alternative to the presence of cold gas . deeper radio observations would be required to decisively settle this issue . our forthcoming _ _ observations of mz10451 should further elucidate the optical properties of the central merging galaxies , and thus help constrain the properties of the merger product itself . these data will also help in establishing the age of tidally stripped stars , and the generation and properties of any intracluster light in this unusual system . a detailed analysis of these and our magellan optical data will be presented in a forthcoming paper . we are very grateful to the referee for insightful and constructive suggestions which improved the presentation of our results . this research has made use of the nasa / ipac extragalactic database ( ned ) . jr acknowledges support provided by the national aeronautics and space administration through chandra postdoctoral fellowship award number pf7 - 80050 issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics and space administration under contract nas8 - 03060 . jsm acknowledges partial support for this study from chandra grant go9 - 0144x . anders , e. , & grevesse , n. 1989 , geochim . cosmochim . acta , 53 , 197 appleton , p. n. , et al . 2004 , , 154 , 147 arp , h. c. , & madore , b. f. 1987 , a catalogue of southern peculiar galaxies and associations 2 volume set , cambridge university press , cambridge , uk bai , l. , rasmussen , j. , mulchaey , j. s. , dariush , a. , raychaudhury , s. , & ponman , t. j. 2010 , , 713 , 637 best , p. n. , von der linden , a. , kauffmann , g. , heckman , t. m. , & kaiser , c. r. 2007 , , 379 , 894 binggeli , b. 1982 , , 107 , 338 boylan - kolchin , m. , ma , c .- p . , & quataert , e. 2006 , , 369 , 1081 brough , s. , couch , w. j. , collins , c. a. , jarrett , t. , burke , d. j. , & mann , r. g. 2008 , , 385 , l103 buote , d. a. 2000 , , 311 , 176 burns , j. o. , hallman , e. j. , gantner , b. , motl , p. m. , & norman , m. l. 2008 , , 675 , 1125 cavagnolo , k. w. , donahue , m. , voit , g. m. , & sun , m. 2008 , , 683 , l107 comastri , a. 2004 , in astrophysics and space science library vol . 308 , supermassive black holes in the distant universe , ed . a. j. barger ( dordrecht : kluwer ) , 245 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , , 115 , 1693 conroy , c. , wechsler , r. h. , & kravtsov , a. v. 2007 , , 668 , 826 cortese , l. , gavazzi , g. , & boselli , a. 2008 , , 390 , 1282 coziol , r. , andernach , h. , caretta , c. a. , alamo - martnez , k. a. , & tago , e. 2009 , , 137 , 4795 de grandi , s. , ettori , s. , longhetti , m. , & molendi , s. 2004 , , 419 , 7 de lucia , g. , & blaizot , j. 2007 , , 375 , 2 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 donghia , e. , sommer - larsen , j. , romeo , a. d. , burkert , a. , pedersen , k. , portinari , l. , & rasmussen , j. 2005 , , 630 , l109 dressler , a. , & shectman , s. a. 1988 , , 95 , 985 dressler , a. , hare , t. , bigelow , b. c. , & osip , d. j. 2006 , , 6269 dubinski , j. 1998 , , 502 , 141 edge , a. c. , & stewart , g. c. 1991 , , 252 , 428 edwards , l. o. v. , hudson , m. j. , balogh , m. l. , & smith , r. j. 2007 , , 379 , 100 faltenbacher , a. , kravtsov , a. v. , nagai , d. , & gottlber , s. 2005 , , 358 , 139 finoguenov , a. , david , l. p. , & ponman , t. j. 2000 , , 544 , 188 finoguenov , a. , davis , d. s. , zimer , m. , & mulchaey , j. s. 2006 , , 646 , 143 fuller , t. m. , west , m. j. , & bridges , t. j. 1999 , , 519 , 22 gil de paz , a. , et al . 2007 , , 173 , 185 gonzalez , a. h. , zabludoff , a. i. , & zaritsky , d. 2003 , , 285 , 67 gregorini , l. , vettolani , g. , de ruiter , h. r. , & parma , p. 1992 , , 95 , 1 hansen , s. m. , sheldon , e. s. , wechsler , r. h. , & koester , b. p. 2009 , , 699 , 1333 hashimoto , y. , henry , j. p. , & bhringer , h. 2008 , , 390 , 1562 hopkins , p. f. , hernquist , l. , cox , t. j. , & kere , d. 2008 , , 175 , 356 jeltema , t. e. , binder , b. , & mulchaey , j. s. 2008 , , 679 , 1162 jeltema , t. e. , mulchaey , j. s. , lubin , l. m. , & fassnacht , c. d. 2007 , , 658 , 865 johnson , r. , ponman , t. j. , & finoguenov , a. 2009 , , 395 , 1287 jones , c. , & forman , w. 1984 , , 276 , 38 jones , l. r. , ponman , t. j. , horton , a. , babul , a. , ebeling , h. , & burke , d. j. 2003 , , 343 , 627 kennicutt , r. c. , jr . 1998 , , 36 , 189 khochfar , s. , & burkert , a. 2003 , , 597 , l117 khosroshahi , h. g. , ponman , t. j. , & jones , l. r. 2007 , , 377 , 595 kitzbichler , m. g. , & white , s. d. m. 2008 , , 391 , 1489 lauer , t. r. 1988 , , 325 , 49 leccardi , a. , & molendi , s. 2008 , , 487 , 461 leccardi , a. , rossetti , m. , & molendi , s. 2010 , , 510 , a82 lin , y .- t . , & mohr , j. j. 2004 , , 617 , 879 liu , f. s. , mao , s. , deng , z. g. , xia , x. y. , & wen , z. l. 2009 , , 396 , 2003 loubser , s. i. , sansom , a. e. , snchez - blzquez , p. , soechting , i. k. , & bromage , g. e. 2008 , , 391 , 1009 mccarthy , i. g. , et al . 2007 , , 376 , 497 mcintosh , d. h. , guo , y. , hertzberg , j. , katz , n. , mo , h. j. , van den bosch , f. c. , & yang , x. 2008 , , 388 , 1537 markevitch , m. , vikhlinin , a. , & mazzotta , p. 2001 , , 562 , l153 mittal , r. , hudson , d. s. , reiprich , t. h. , & clarke , t. 2009 , , 501 , 835 mulchaey , j. s. , davis , d. s. , mushotzky , r. f. , & burstein , d. 2003 , , 145 , 39 mulchaey , j. s. , lubin , l. m. , fassnacht , c. , rosati , p. , & jeltema , t. e. 2006 , , 646 , 133 murante , g. , giovalli , m. , gerhard , o. , arnaboldi , m. , borgani , s. , & dolag , k. 2007 , , 377 , 2 nipoti , c. , stiavelli , m. , ciotti , l. , treu , t. , & rosati , p. 2003 , , 344 , 748 osmond , j. p. f. , & ponman , t. j. 2004 , , 350 , 1511 ostriker , j. p. , & hausman , m. a. 1977 , , 217 , l125 osullivan , e. , forbes , d. a. , & ponman , t. j. 2001 , , 328 , 461 osullivan , e. , ponman , t. j. , & collins , r. s. 2003 , , 340 , 1375 paul , s. , iapichino , l. , miniati , f. , bagchi , j. , & mannheim , k. 2010 , , submitted ( arxiv:1001.1170 ) pimbblet , k. a. 2008 , publications of the astronomical society of australia , 25 , 176 poole , g. b. , fardal , m. a. , babul , a. , mccarthy , i. g. , quinn , t. , & wadsley , j. 2006 , , 373 , 881 quintana , h. , & lawrie , d. g. 1982 , , 87 , 1 rafferty , d. a. , mcnamara , b. r. , & nulsen , p. e. j. 2008 , , 687 , 899 rasmussen , j. , & ponman , t. j. 2007 , , 380 , 1554 rasmussen , j. , & ponman , t. j. 2009 , , 399 , 239 rasmussen , j. , ponman , t. j. , mulchaey , j. s. , miles , t. a. , & raychaudhury , s. 2006 , , 373 , 653 rebusco , p. , churazov , e. , bhringer , h. , & forman , w. 2005 , , 359 , 1041 rines , k. , finn , r. , & vikhlinin , a. 2007 , , 665 , l9 rogers , b. , ferreras , i. , kaviraj , s. , pasquali , a. , & sarzi , m. 2009 , , 399 , 2172 sanderson , a. j. r. , osullivan , e. , & ponman , t. j. 2009 , , 395 , 764 sanderson , a. j. r. , ponman , t. j. , & osullivan , e. 2006 , , 372 , 1496 stanghellini , l. , gonzlez - garca , a. c. , & manchado , a. 2006 , , 644 , 843 sun , m. , jones , c. , forman , w. , vikhlinin , a. , donahue , m. , & voit , m. 2007 , , 657 , 197 sun , m. , voit , g. m. , donahue , m. , jones , c. , forman , w. , & vikhlinin , a. 2009 , , 693 , 1142 sutherland , r. s. , & dopita , m. a. 1993 , , 88 , 253 tran , k .- v . h. , moustakas , j. , gonzalez , a. h. , bai , l. , zaritsky , d. , & kautsch , s. j. 2008 , , 683 , l17 tremaine , s. 1990 , in dynamics and interactions of galaxies , ed . r. wielen ( berlin : springer ) , 394 tremaine , s. d. , & richstone , d. o. 1977 , , 212 , 311 vale , a. , & ostriker , j. p. 2008 , , 383 , 355 west , m. j. 1994 , , 268 , 79 yamada , t. , koyama , y. , nakata , f. , kajisawa , m. , tanaka , i. , kodama , t. , okamura , s. , & de propris , r. 2002 , , 577 , l89 yun , m. s. , reddy , n. a. , & condon , j. j. 2001 , , 554 , 803 zabludoff , a. i. , & mulchaey , j. s. 1998 , , 496 , 39 zabludoff , a. i. , & mulchaey , j. s. 2000 , , 539 , 136 zabludoff , a. i. , huchra , j. p. , & geller , m. j. 1990 , , 74 , 1 zaritsky , d. , gonzalez , a. h. , & zabludoff , a. i. 2004 , , 613 , l93	the central dominant galaxies in galaxy clusters constitute the most massive and luminous galaxies in the universe . despite this , the formation of these brightest cluster galaxies ( bcgs ) and the impact of this on the surrounding cluster environment remain poorly understood . here we present multi - wavelength observations of the nearby poor x - ray cluster mz10451 , in which both processes can be studied in unprecedented detail . _ chandra _ observations of the intracluster medium ( icm ) in the cluster core , which harbors two optically bright early - type galaxies in the process of merging , show that the system has retained a cool core and a central metal excess . this suggests that any merger induced icm heating and mixing remain modest at this stage . tidally stripped stars seen around either galaxy likely represent an emerging intracluster light component , and the central icm abundance enhancement may have a prominent contribution from _ in situ _ enrichment provided by these stars . the smaller of the merging galaxies shows evidence for having retained a hot gas halo , along with tentative evidence for some obscured star formation , suggesting that not all bcg major mergers at low redshift are completely dissipationless . both galaxies are slightly offset from the peak of the icm emission , with all three lying on an axis that roughly coincides with the large - scale elongation of the icm . our data are consistent with a picture in which central bcgs are built up by mergers close to the cluster core , by galaxies infalling on radial orbits aligned with the cosmological filaments feeding the cluster .
graphitic systems are used for a wide variety of industrial applications , ranging from lubricant and refractory materials to neutron moderators in nuclear fission reactors@xcite and plasma shields in the next generation of fusion reactors@xcite . the recent realization of graphene@xcite ( single graphitic layer ) and the discovery of its unusual electronic properties@xcite have raised the interest on flake graphitic systems as a route to produce graphene samples of high quality and in large scale@xcite . despite the technological and scientific importance of graphitic systems , the knowledge of their elastic properties is unexpectedly poor and new insights are needed . the values of the elastic constants describe the mechanical behaviour@xcite and are decisive in engineering design to avoid material failure . in layered materials , they are even more important for the thermodynamic properties due to a low - lying branch of acoustic vibrations , the bending modes , predicted by lifshitz@xcite over fifty years ago . here we show that the shear elastic constant @xmath0 affects the mechanism of exfoliation that is relevant for the production of graphene . the most reliable experimental studies of the elastic constants have been obtained by inelastic x - ray scattering@xcite and ultrasonic , sonic resonance , and static test methods@xcite . the sample used in the first study@xcite was single - crystalline kish graphite , characterized by an extraordinary high degree of ordering , the closest approximation to the perfect ab stacking graphite ( hex - g ) . the second study@xcite was done using highly oriented pyrolitic graphite , the closest approximation to turbostratic graphite ( turbo - g ) where the graphitic layers are randomly oriented around the c - axis . except for @xmath0 and @xmath1 , both studies are in agreement within the experimental uncertainties . the @xmath1 value in turbo - g was determined only by the less accurate static test method and it may be affected by errors . conversely , @xmath0 in turbo - g was determined from the sound velocity and its value ranges between 0.18 - 0.35 gpa@xcite , one order of magnitude lower than 5.03 gpa found in hex - g@xcite . the discrepancy on @xmath0 is attributed to the existence of mobile basal dislocations@xcite . after neutron irradiation , the elastic constant @xmath0 increases by up to an order of magnitude@xcite , suggesting that interstitial defects@xcite could pin the dislocation motions , whence the intrinsic value of @xmath0 is measured . a principal difficulty with this explanation is that interstitial atoms inevitably increase the shear resistance between graphitic layers and therefore they may increase the @xmath0 value by themselves . the aim of this study is to investigate from first principles the elastic constants of graphitic systems with respect to the stacking misorientations between layers , and to describe the key role of the shear elastic constant @xmath0 on the bending modes ( thermal property ) and mechanical stability . we show that stacking misorientations greatly affect @xmath0 and that graphitic systems with perfect ( hex - g ) and random ( turbo - g ) stacking should be considered as two distinct materials described by their own elastic and thermodynamic properties . this paper is structured as follow . in sec . [ method ] , we give a brief summary of the theoretical methods and a discussion of the lcao - s@xmath2+vdw formalism to include the long - range van der waals ( vdw ) interactions . the consequences of shear elastic constant @xmath0 on the bending modes and mechanical stability are shown in sec . [ bending_modes ] and [ mechanical_stability ] , respectively . the elastic constants in the case of high - symmetric systems ( hexagonal , orthorhombic , rhombohedral and aa hexagonal stackings ) and for graphitic layers randomly oriented around the c - axis ( turbostratic stacking ) are presented in sec . [ ec_high - symmetric ] and [ ec_turbostratic ] , respectively . finally , we summarize and comment on our results ( sec . [ conclusion ] ) . all the calculations are performed using density functional theory , within the local density approximation scheme ( lda ) , norm - conserving pseudopotentials@xcite and plane waves with cut - off energy of 150 ry ( abinit package)@xcite . the k - point mesh was chosen so that the average density corresponds approximately to a 32x32x16 mesh for hex - g . energies were converged within 0.05 mev / atom , and elastic constants within 0.5% . for large supercells ( turbo - g with more than 50 atoms ) we have used localized basis - set composed of gaussian orbitals ( aimpro code)@xcite . the elastic constants calculated by the two lda codes are in agreement within 3% or better . the choice of lda is not fortuitous@xcite and it was dictated by test calculations using the generalized - gradient approximation ( gga ) within the perdew - burke - ernzerhof scheme@xcite . according to gga , the distance between graphitic layers is far too large ( 4.2 ) , resulting in a negligible interlayer binding energy and almost vanishing out - of - plane elastic constants ( @xmath0 , @xmath3 ) . for these reasons we have dismissed the use of gga from this study@xcite . even though lda yields accurate equilibrium distances , due to energetical error compensations , the long - range van der waals ( vdw ) and more generally the weak contribution to the out - of - plane interactions is not well described@xcite . in order to check the importance of these effects on @xmath0 and @xmath3 , we have used the lcao - s@xmath2+vdw formalism to include these specific interactions within lda@xcite ( using fireball code)@xcite . this formalism takes into account two major contributions . the first one , which we call weak chemical interaction , is a repulsive energy originating from the overlaps of electronic densities between the weakly interacting subsystems . even though the overlaps are rather small , this energy is not negligible . this contribution is evaluated proceeding to a second order expansion of the electronic wavefunctions with respect to the overlaps . the second contribution , which is the vdw itself , originating from charge fluctuations , can be seen as the interaction between electronic dipoles . in the frame of the dipolar approximation , we use a second order perturbation theory to describe this contribution . this method has originally been tested with success on standard graphene - graphene interaction , and more recently on a wide range of graphitic materials@xcite . in its current stage , the analysis of the internal stresses is not implemented yet , thus the elastic constants that change the in - plane bond lengths ( @xmath4 , @xmath5 , @xmath1 ) are overestimated by up to 15% . however , for @xmath0 and @xmath3 that describe the weak interaction between layers , the internal stresses are negligible , and the calculated values are expected to be very accurate . the elastic constants are determined using two different approaches . the first one uses the response - function , implemented in the abinit code , to calculate the second derivative of the total energy with respect to the strains . the second approach uses the elastic energy density@xcite . for each elastic constant we have applied 21 strain components @xmath6 to the equilibrium crystal structures ( @xmath6 were typically ranging between 0.01 with increments 0.001 ) . the elastic constants are found by fitting the calculated energies to a polynomial function in the strains . both approaches yield results in agreement within 0.2% . the bending modes are atomic vibrations that can be excited even at low temperature and strongly influence the thermal properties of layered materials . in 1952 , lifshitz@xcite obtained the following dispersion law for the out - of - plane acoustic mode @xmath7 : @xmath8 where @xmath9 is the density , @xmath10 is the interlayer distance , @xmath11 are the wave vectors and @xmath12 describes the intralayer forces characterizing the bending rigidity ( 1.1 ev)@xcite . ( color online ) transversal acoustic ( bending ) mode . the bending changes the local stacking between graphitic layers . the boxes ( a - d ) show regions with different slopes and stackings . the shear @xmath13 gives the deviations from ab phase ( perfect stacking ) . , width=321 ] the small value of @xmath0 ( characteristic of graphitic systems , see table [ table : elastic_constants ] ) leads to a predominant contribution of the transversal bending modes ( @xmath14 ) in the phonon dispersion curves . these modes are sinusoidal displacements that propagate along the planes and change the local stacking between layers ( see fig . [ figure1 ] ) . for nearly flat planes the shear stacking @xmath13 is almost zero , whereas , for positive ( or negative ) slope @xmath13 becomes positive ( or negative ) . using trigonometric considerations , it can be shown that the maximum value of @xmath13 ( see boxes in fig . [ figure1 ] ) is given by : @xmath15 where @xmath16 is the amplitude , @xmath17 is the wavelength and @xmath10 is the interlayer distance . the crystal resistance to the stacking shear@xcite is proportional to @xmath18 , and by lowering @xmath0 more bending modes can be excited at lower temperature . in the limit of @xmath19 , graphitic systems approach the graphene behaviour , where indeed bending modes ( or ripples ) are always present@xcite . by imposing the elastic strain energy as positively definite@xcite , the stability conditions are given by : @xmath20 note that @xmath1 does not affect the stability : a positive ( or negative ) value means that under in - plane compression the out - of - plane distance tends to expand ( or contract ) . the elastic constants @xmath4 , @xmath5 describe in - plane deformations and they possess the highest values due to the strong sp@xmath2 bonding interactions within the graphene planes . the elastic constant @xmath3 describes out - of - plane compression or expansion and it has always a positive value for perfect and stacking misorientations ( see table [ table : elastic_constants ] ) . the elastic constant @xmath0 corresponds to a shear between graphene layers . due to the weak interaction between planes , the @xmath0 value is the lowest and can be positive or negative depending on the stacking misorientations ( see table [ table : elastic_constants ] ) . the latter elastic constant is the only one that can break the mechanical stability condition ( _ i.e. _ @xmath21 ) . by imposing a translation vector between graphitic layers we found four high - symmetric stackings that correspond to stationary points on the stacking - fault energy surface@xcite . to calculate the stacking - fault energy surface ( for graphitic system is often called corrugation energy surface)@xcite , we have used 16-atoms unit cell model of eight layers in ab sequence in which the stacking at the unit cell boundary is changed by imposing a shear displacement . this represents an intrinsic stacking fault between layers at the unit cell boundaries whereas the others remains stacked in the ab sequence . the multiple layers repeat along the c - axis ( 8 layers ) makes negligible self - interaction of the intrinsic fault@xcite . the stationary points , indicated with square and circle symbols in fig . [ figure2]a , correspond to the four high symmetric structures ( hexagonal , rhombohedral , orthorhombic and aa hexagonal graphite , see fig . [ figure2]b ) . the in - plane lattice parameter was 2.45 very close to the experimental value 2.463@xcite , with no significant changes among all the graphitic structures . the origin in fig . [ figure2]a corresponds to hex - g ( black square symbol ) , the global minimum of the energy surface with an interlayer separation of 3.34 ( experimental value 3.356 )@xcite . with the exception of the @xmath3 value ( 29 gpa ) , the calculated elastic constants ( given in table [ table : elastic_constants ] ) are in agreement within the experimental uncertainties found in hex - g@xcite . using the lcao - s@xmath2+vdw formalism , the out - of - plane elastic constant @xmath3 becomes 42 gpa , very close to the experimental value 38.7@xmath227 gpa@xcite . ( a ) stacking - fault energy surface ( also called corrugation energy surface ) . the square and circle symbols indicate the stationary points corresponding to the following high - symmetric structures ; ( b ) the hexagonal , orthorhombic , rhombohedral and aa hexagonal stackings viewed perpendicular ( above ) , parallel ( below ) to the c - axis . the energies of these structures with respect to hex - g are 1.66 mev / atom for ortho - g , 0.10 mev / atom for the rhom - g and 9.29 mev / atom for aa hex - g ( see text ) . , width=321 ] the local minimum at @xmath23@xmath24@xmath25@xmath26 ( black circle symbol ) represents the rhombohedral stacking ( rhombo - g ) . this structure possesses the same interlayer separation and elastic constants of hex - g ( the differences are beyond the accuracy of the calculations ) with formation energy of 0.10 mev / atom . this very small energy explains why the rhombohedral phase is usually 5 - 15% intermixed with the perfect hexagonal one in natural graphitic flakes . the saddle point at @xmath27@xmath24@xmath25@xmath26 ( grey square symbol ) represents the orthorhombic stacking ( ortho - g ) . the interlayer separation of the primitive unit cell is 3.37 with formation energy of 1.66 mev / atom . the latter energy represents the lowest barrier that has to be overcome during the shearing process from an ideal configuration to another equivalent one . the @xmath0 values are found to range between -2.7 gpa , for shearing along @xmath24@xmath25@xmath26 , and 7.7 gpa , for shearing along the @xmath24@xmath28@xmath26 axis . no significant changes are found for the other elastic constants ( see table [ table : elastic_constants ] ) . although unstable ( due to the negative @xmath0 ) , this structure acts as an intermediate phase during the transformation from graphite to diamond@xcite . the global maximum at @xmath29@xmath24@xmath25@xmath26 ( grey circle symbol ) corresponds to aa hexagonal stacking ( aa hex - g ) . this phase has the largest interlayer separation ( 3.60 ) and the highest formation energy ( 9.29 mev / atom ) . its elastic constants are smaller than in hex - g , and @xmath0 in particular , becomes negative ( -3.8 gpa ) breaking the stability conditions . even though this structure is highly unstable , a recent study has suggested that screw dislocations locally encourage this stacking@xcite . in the following section we describe the elastic constants for graphitic layers randomly oriented around the c - axis . the modelling of turbostratic staking is challenging since the incommensurate nature of these stackings must combine with the finite - size constraint required by calculations . to overcome this difficulty we used the method proposed by kolmogorov and crespi@xcite in which a layer with supercell basis vector @xmath30 becomes commensurate with a second layer for a specific rotation angle of : @xmath31 \qquad \textrm{with}\;\ ; n > m\ ] ] ( color online ) ( a , b ) representation of accidental commensuration . a supercell of vector @xmath32 ( blue colour ) becomes commensurate when rotated by an angle @xmath33 with respect to the starting supercell ( as we increase @xmath34 the supercell surface and number of atoms rapidly increase ) . ( c ) stacking fault energy surface ( or corrugation energy surface ) of a graphene bilayer with supercell vector ( 2,1 ) . the respective elastic constants are calculated on the local minima of the energy surface . notice that the energy surface becomes much flatter leading to a reduction of @xmath0 . , width=283 ] figure [ figure3]a , b shows the case of the two smaller supercell with basis vector @xmath35 and @xmath36 ( corresponding to 14 , 26 atoms / layer and rotation angles of 38.21@xmath37 , 32.20@xmath37 , respectively ) . figure [ figure3]c shows the stacking energy surface for a bilayer supercell of basis vector @xmath35 . to better describe the complete misorientation of turbostratic stacking we have increased the number of layers along the c - axis . each layer was rotated with respect to each other and randomly translated along the basal plane ( see fig . [ figure4 ] ) . the rotation angles used are 15 values ranging from 6.01@xmath37 to 53.99@xmath37 . the smallest supercell contains 28 atoms with 2 layers rotated with respect to each other by 38.21@xmath37 , whereas the largest one contains 456 atoms with 12 layers and rotation angle of 46.83@xmath37 . for each model we carried out extensive structural optimizations starting from different translation vectors along the basal plane . the corrugation energy is about one order of magnitude lower than commensurate structure ( see fig . [ figure2]a ) with a maximum value of 0.9 mev / atom ( see fig . [ figure3]c ) . as we increased the size of the supercell over the basal plane we found that the average corrugation energy tends to decrease up to 20% for the largest size ( basis vector 8,3 with 194 atoms / layer ) . extrapolating these results in the ideal case of infinite layers , we suggest corrugation energy virtually flat with layers mutual independent . for all the supercell studied the in - plane lattice parameters remain almost equal to the value found in hex - g , whereas the interlayer distances were on average slightly larger , [email protected] , with formations energies of [email protected] mev / atom . an example of 5-layers turbostratic stacking ( supercell basis vector 2,1 ) . each layer is stacked along the c - axis , rotated with respect to each other with an angle of 38.21@xmath37 and randomly translated along the basal plane . , width=321 ] the interlayer binding energy between graphitic layers ( _ i.e. _ exfoliation energy ) was 21@xmath221 mev / atom ( 70@xmath224 mev / atom with vdw ) , a value slightly lower than the 24 mev / atom found in hex - g ( 80 mev / atom with vdw ) . note that lda values yield to a binding energy within a factor of 2 - 3 with respect to lcao - s@xmath2+vdw formalism and experiment values ( 43 mev / atom found in heat - of - wetting experiment@xcite , 35@xmath2210 mev / atom found by analyzing tem images of twisted collapsed nanotubes@xcite , and 52@xmath225 mev / atom by studying thermal desorption of polyaromatic hydrocarbons@xcite ) . in table [ table : elastic_constants ] we report the calculated values of the elastic constants . with the exception of @xmath0 , these values hardly change among all the studied turbostratic stackings with no clear dependence on the rotation angles and number of layers . the elastic constants @xmath4 , @xmath5 slightly decrease by about 3% with respect to hex - g , remaining within the experimental uncertainties found in turbostratic samples@xcite . as previously found in hex - g , lda calculations underestimate the @xmath3 value ( 27@xmath222 gpa ) with respect to the vdw correction ( 36@xmath221 gpa ) and the experimental value of 36.5@xmath221gpa @xcite . conversely , we found @[email protected] gpa in disagreement with the experimental value 15@xmath225 gpa@xcite . the latter value was only indirectly obtained as a function of the other elastic constants by the less accurate static test method . this method requires larger strains than ultrasonic experiments and non - linear behaviour of stress - strain curve may affect the measured value . furthermore , the linear bulk modulus @xmath38 calculated from these elastic constants is far too large ( 2080 gpa ) , almost double than the one found in diamond ( 1326 gpa)@xcite . ( color online ) in - plane lattice parameters _ vs. _ pressure . the solid line represents the results found for hex - g and rhombo - g ( perfect and rhombohedral stackings ) . the dashed line shows the results found here for turbostratic graphite . for comparison , the experimental results are also plotted . , width=313 ] the linear bulk modulus @xmath38 describes the variation of the lattice parameter @xmath39 as a function of the hydrostatic pressure@xcite and it is given by : @xmath40 this modulus is strongly weighted by @xmath1 . for example , if we use the measured value found in hex - g 0@xmath223 gpa ( instead of 15@xmath225 gpa ) , @xmath38 becomes 1240 gpa ( instead of 2080 gpa ) . several x - ray studies have measured the linear bulk modulus @xmath38 . in these experiments powder samples were prepared by grinding different types of graphitic materials , ranging from well crystallized to poorly crystallized grains@xcite . the good agreement between the experiments ( see fig . [ figure5 ] ) indicates that the linear bulk modulus @xmath38 does not strongly depend on the stacking order with a measured value of 1250@xmath2270 gpa@xcite . by including the latter modulus in the eq . [ eq : linear_bulk_modulus ] , the elastic constant @xmath1 becomes 0.3 gpa . therefore , we conclude that the @xmath1 value does not significantly change between turbo - g and hex - g and we propose that the same value 0@xmath223 gpa should be appropriate also for turbostratic stacking . the @xmath0 represents the second derivative of the total energy as a function of shear displacement over the basal plane . as the corrugation energy of perfect ab stacking is much higher than turbostratic we are expecting a corresponding lower value of the shear elastic constant . we have found that @xmath0 tends to decrease as a function of the supercell area over the basal plane ( see fig . [ figure6 ] ) and is independent on the relative rotation angles and number of layers along the c - axis ( the respective average corrugation energies and therefore the @xmath0 values are also independent on the relative rotation angles and number of layers ) . ( color online ) the elastic constants @xmath0 and @xmath3 as a function of the commensurate area over the basal plane in turbostratic stacking . the @xmath0 value tends to decrease with respect to the supercell size , whereas @xmath3 is not quantitatively affected . , width=287 ] for commensurate structures of finite area ranging between 36 - 563 @xmath41 the respective @xmath0 values are within 0.16 - 0.31 gpa , one order of magnitude lower than hex - g ( 4.5 gpa ) and close to the experimental measures ranging from 0.18 to 0.35 gpa@xcite . the vdw correction does not significantly affect the @xmath0 values with respect to lda ( see table [ table : elastic_constants ] ) , suggesting that the variations in energy under interlayer shears are nearly identical between these two approximations . as we increase the sizes of the commensurate structures towards the ideal case of infinite layers ( incommensurate ) the @xmath0 values tend to zero with corrugation energy virtually flat . these results indicate that turbostratic stacking possesses the lowest friction among all the graphitic materials bearing great potential applications in nano - mechanical systems@xcite . [ cols="<,<,<,>,<,<,<,>,^,^,^,>,^,<,^,>,^,<,^,>,>,<,<,<,>,>,<,^,^ , > , < " , ] to summarize , we have discussed the importance of @xmath0 as the main parameter that restrains the bending modes and controls the mechanical stability of layered materials . using advanced _ ab - initio _ method , which includes vdw interactions , we have provided the first complete description of the elastic constants in graphitic systems . the lower exfoliation energy ( 3 - 8 mev / atom ) and the lower @xmath0 ( at least one order of magnitude ) found in turbostratic stacking suggest that the exfoliation mechanism , relevant for the production of graphene , should be easier for graphite flakes with random stacking . our results indicate that turbostratic graphitic systems possess the lowest friction among all the graphitic stackings . it would be interesting to check these predictions experimentally . the authors thank the hpc - europa project for financial and computer supports . computational resources have also been provided by the universit catholique de louvain on the lemaitre and green computers of the cism . c.c . acknowledges financial support from the f.r.s .- fnrs of belgium ; g.s . from jsps and grant - in - aid for scientific research . evans te , moyer ra , burrell kh , fenstermacher me , joseph i , leonard aw , et al . edge stability and transport control with resonant magnetic perturbations in collisionless tokamak plasmas . nature phys . 2006 ; 2:419 - 423 . scaffidi - argentina f , safronov v , arkhipov i , arkhipov n , bakhtin v , barsuk v , et al . erosion mechanisms and products in graphite targets under simulated disruption conditions . j. nucl . mater . 2000;283 - 287(2):1111 - 1115 . mohiuddin tmg , lombardo a , nair rr , bonetti a , savini g , jalil r et al . uniaxial strain in graphene by raman spectroscopy : g peak splitting , grneisen parameters , and sample orientation . b 2009;79(20):205433 . it is clear that gga ( leading to a distance of 4.2 and energy of 2 - 3 mev / atom ) seriously underestimates the interlayer binding energy and is in strong disagreement for the geometry . thus , even if lda is unable to describe the dispersion interactions , it is a better choice than the gga as far as the geometry and energy differences are concerned . dappe yj , basanta ma , flores f , ortega j. weak chemical interaction and van der waals forces between graphene layers : a combined density functional and intermolecular perturbation theory approach . b 2006;74(20):205434 . yagi t , utsumi w , yamakata m , kikegawa t , shimomura o. high - pressure in situ x - ray - diffraction study of the phase transformation from graphite to hexagonal diamond at room temperature . b 1992;46(10):6031 - 6039 . zhao yx , spain il . x - ray diffraction data for graphite to 20 gpa . b 1989;40(2):993 - 997 ; to estimate @xmath38 they inadvertently used the planar bulk modulus instead of using the linear one ( see schreiber et al .	the thermodynamic and mechanical properties of graphitic systems are strongly dependent on the shear elastic constant @xmath0 . using state - of - the - art density functional calculations , we provide the first complete determination of their elastic constants and exfoliation energies . we show that stacking misorientations lead to a severe lowering of @xmath0 of at least one order of magnitude . the lower exfoliation energy and the lower @xmath0 ( more bending modes ) suggest that flakes with random stacking should be easier to exfoliate than the ones with perfect or rhombohedral stacking . we also predict ultralow friction behaviour in turbostratic graphitic systems .
let @xmath5 be an abelian variety defined over a finite field @xmath6 , and @xmath9 a prime number dividing the order of the group @xmath10 . then the _ embedding degree _ of @xmath5 with respect to @xmath7 is the degree of the field extension @xmath11 obtained by adjoining a primitive @xmath7-th root of unity @xmath12 to @xmath6 . the embedding degree is a natural notion in pairing - based cryptography , where @xmath5 is taken to be the jacobian of a curve defined over @xmath6 . in this case , @xmath5 is principally polarized and we have the non - degenerate _ weil pairing _ @xmath13\times a[r ] \longrightarrow \mu_r\ ] ] on the subgroup scheme @xmath14 $ ] of @xmath7-torsion points of @xmath5 with values in the @xmath7-th roots of unity . if @xmath6 contains @xmath12 , we also have the non - trivial _ tate pairing _ @xmath15({{\mathbf f } } ) \times a({{\mathbf f}})/ra({{\mathbf f } } ) \to \mathbf{f}^/({{\mathbf f}}^)^r.\ ] ] the weil and tate pairings can be used to ` embed ' @xmath7-torsion subgroups of @xmath10 into the multiplicative group @xmath16 , and thus the discrete logarithm problem in @xmath17 $ ] can be ` reduced ' to the same problem in @xmath16 @xcite . in pairing - based cryptographic protocols @xcite , one chooses the prime @xmath7 and the embedding degree @xmath8 such that the discrete logarithm problems in @xmath17 $ ] and @xmath16 are computationally infeasible , and of roughly equal difficulty . this means that @xmath7 is typically large , whereas @xmath8 is small . jacobians of curves meeting such requirements are often said to be _ pairing - friendly_. if @xmath6 has order @xmath3 , the embedding degree @xmath18 $ ] is simply the multiplicative order of @xmath3 in @xmath19 . as ` most ' elements in @xmath19 have large order , the embedding degree of @xmath5 with respect to a large prime divisor @xmath7 of @xmath20 will usually be of the same size as @xmath7 , and @xmath5 will not be pairing - friendly . one is therefore led to the question of how to efficiently construct @xmath5 and @xmath6 such that @xmath10 has a ( large ) prime factor @xmath7 and the embedding degree of @xmath5 with respect to @xmath7 has a prescribed ( small ) value @xmath8 . the current paper addresses this question on two levels : the _ existence _ and the actual _ construction _ of @xmath5 and @xmath6 . section [ s : weilnumbers ] focuses on the question whether , for given @xmath7 and @xmath8 , there exist abelian varieties @xmath5 that are defined over a finite field @xmath6 , have an @xmath6-rational point of order @xmath7 , and have embedding degree @xmath8 with respect to @xmath7 . we consider only abelian varieties @xmath5 that are _ simple _ , that is , not isogenous ( over @xmath6 ) to a product of lower - dimensional varieties , as we can always reduce to this case . by honda - tate theory @xcite , isogeny classes of simple abelian varieties @xmath5 over the field @xmath6 of @xmath3 elements are in one - to - one correspondence with @xmath21-conjugacy classes of _ @xmath3-weil numbers _ , which are algebraic integers @xmath22 with the property that all embeddings of @xmath22 into @xmath23 have absolute value @xmath24 . this correspondence is given by the map sending @xmath5 to its @xmath3-th power frobenius endomorphism @xmath22 inside the number field @xmath25 . the existence of abelian varieties with the properties we want is thus tantamount to the existence of suitable weil numbers . our main result , algorithm [ alg : construct - pi ] , constructs suitable @xmath3-weil numbers @xmath22 in a given _ cm - field _ it exhibits @xmath22 as a _ type norm _ of an element in a _ reflex field _ of @xmath0 satisfying certain congruences modulo @xmath7 . the abelian varieties @xmath5 in the isogeny classes over @xmath6 that correspond to these weil numbers have an @xmath6-rational point of order @xmath7 and embedding degree @xmath8 with respect to @xmath7 . moreover , they are _ ordinary _ , i.e. , @xmath26=p^g$ ] , where @xmath27 is the characteristic of @xmath6 . theorem [ thm : ispoltime ] shows that for fixed @xmath0 , the expected run time of our algorithm is heuristically polynomial in @xmath28 . for an abelian variety of dimension @xmath29 over the field @xmath6 of @xmath3 elements , the group @xmath10 has roughly @xmath30 elements , and one compares this size to @xmath7 by setting @xmath31 in cryptographic terms , @xmath32 measures the ratio of a pairing - based system s required bandwidth to its security level , so small @xmath32-values are desirable . _ supersingular _ abelian varieties can achieve @xmath32-values close to @xmath33 , but their embedding degrees are limited to a few values that are too small to be practical @xcite . theorem [ thm : heur ] discusses the distribution of the ( larger ) @xmath32-values we obtain . in section [ s : construct ] , we address the issue of the actual construction of abelian varieties corresponding to the weil numbers found by our algorithm . this is accomplished via the construction in characteristic zero of the abelian varieties having cm by the ring of integers @xmath34 of @xmath0 , a hard problem that is far from being algorithmically solved . we discuss the elliptic case @xmath35 , for which reasonable algorithms exist , and the case @xmath36 , for which such algorithms are still in their infancy . for genus @xmath37 , we restrict attention to a few families of curves that we can handle at this point . our final section [ s : examples ] provides numerical examples . let @xmath6 be a field of @xmath3 elements , @xmath5 a @xmath29-dimensional simple abelian variety over @xmath6 , and @xmath38 the number field generated by the frobenius endomorphism @xmath22 . then @xmath22 is a _ @xmath3-weil number _ in @xmath0 : an algebraic integer with the property that all of its embeddings in @xmath39 have complex absolute value @xmath24 . the @xmath3-weil number @xmath22 determines the group order of @xmath10 : the @xmath6-rational points of @xmath5 form the kernel of the endomorphism @xmath40 , and in the case where @xmath41 is the full endomorphism algebra @xmath42 we have @xmath43 in the case @xmath44 we will focus on , @xmath0 is a _ cm - field _ of degree @xmath45 as in ( * ? ? ? * section 1 ) , i.e. , a totally complex quadratic extension of a totally real subfield @xmath46 . [ p : exist - embed ] let @xmath5 , @xmath6 and @xmath22 be as above , and assume @xmath41 equals @xmath47 . let @xmath8 be a positive integer , @xmath48 the @xmath8-th cyclotomic polynomial , and @xmath49 a prime number . if we have @xmath50 then @xmath5 has embedding degree @xmath8 with respect to @xmath7 . the first condition tells us that @xmath7 divides @xmath20 , the second that the order of @xmath51 in @xmath19 , which is the embedding degree of @xmath5 with respect to @xmath7 , equals @xmath8 . by honda - tate theory @xcite , all @xmath3-weil numbers arise as frobenius elements of abelian varieties over @xmath6 . thus , we can prove the _ existence _ of an abelian variety @xmath5 as in proposition [ p : exist - embed ] by exhibiting a @xmath3-weil number @xmath52 as in that proposition . the following lemma states what we need . [ l : honda - tate ] let @xmath22 be a @xmath3-weil number and @xmath6 be the field of @xmath3 elements . then there exists a unique isogeny class of simple abelian varieties @xmath53 with frobenius @xmath22 . if @xmath41 is totally imaginary of degree @xmath45 and @xmath3 is prime , then such @xmath5 have dimension @xmath29 , and @xmath0 is the full endomorphism algebra @xmath54 . if furthermore @xmath3 is unramified in @xmath0 , then @xmath5 is ordinary . the main theorem of @xcite yields existence and uniqueness , and shows that @xmath55 is a central simple algebra over @xmath41 satisfying @xmath56^{\frac{1}{2}}[k:{\mathbf{q}}].\ ] ] for @xmath0 totally imaginary of degree @xmath45 and @xmath3 prime , waterhouse ( * ? ? ? * theorem 6.1 ) shows that we have @xmath57 and @xmath58 . by ( 7.1 ) , @xmath5 is ordinary if and only if @xmath59 is prime to @xmath60 in @xmath34 . thus if @xmath5 is not ordinary , the ideals @xmath61 and @xmath62 have a common divisor @xmath63 with @xmath64 , so @xmath3 ramifies in @xmath0 . [ cycliccase ] our general construction is motivated by the case where @xmath0 is a galois cm - field of degree @xmath45 , with cyclic galois group generated by @xmath65 . here @xmath66 is complex conjugation , so we can construct an element @xmath4 satisfying @xmath67 by choosing any @xmath68 and letting @xmath69 . for such @xmath22 , we have @xmath70 . if @xmath71 is a prime @xmath3 , then @xmath22 is a @xmath3-weil number in @xmath0 . now we wish to impose the conditions of proposition [ p : exist - embed ] on @xmath22 . let @xmath7 be a rational prime that splits completely in @xmath0 , and @xmath72 a prime of @xmath34 over @xmath7 . for @xmath73 , put @xmath74 ; then the factorization of @xmath7 in @xmath34 is @xmath75 . if @xmath76 is the residue class of @xmath77 modulo @xmath78 , then @xmath79 modulo @xmath72 is also @xmath80 , so the residue class of @xmath22 modulo @xmath72 is @xmath81 . furthermore , the residue class of @xmath82 modulo @xmath72 is @xmath83 . if we choose @xmath77 to satisfy @xmath84 we find @xmath85 and thus @xmath86 . by choosing @xmath77 such that in addition @xmath87 is a primitive @xmath8-th root of unity in @xmath88 , we guarantee that @xmath89 is a primitive @xmath8-th root of unity modulo @xmath7 . thus we can try to find a weil number as in proposition [ p : exist - embed ] by picking residue classes @xmath90 for @xmath73 meeting the two conditions above , computing some ` small ' lift @xmath91 with @xmath92 , and testing whether @xmath69 has prime norm . as numbers of moderate size have a high probability of being prime by the prime number theorem , a small number of choices @xmath93 should suffice . there are @xmath94 possible choices for @xmath95 , where @xmath96 is the euler totient function , so for @xmath97 and large @xmath7 we are very likely to succeed . for @xmath35 , there are only a few choices @xmath98 , but one can try various lifts and thus recover what is known as the cocks - pinch algorithm ( * ? ? ? * theorem 4.1 ) for finding pairing - friendly elliptic curves . for arbitrary cm - fields @xmath0 , the appropriate generalization of the map @xmath99 in example [ cycliccase ] is provided by the _ type norm_. a _ cm - type _ of a cm - field @xmath0 of degree @xmath45 is a set @xmath100 of embeddings of @xmath0 into its normal closure @xmath101 such that @xmath102 is the complete set of embeddings of @xmath0 into @xmath101 . the _ type norm _ @xmath103 with respect to @xmath104 is the map @xmath105 which clearly satisfies @xmath106 if @xmath0 is not galois , the type norm @xmath107 does not map @xmath0 to itself , but to its _ reflex field _ @xmath108 with respect to @xmath104 . to end up in @xmath0 , we can however take the type norm with respect to the _ reflex type _ @xmath109 , which we will define now ( cf . * section 8) ) . let @xmath110 be the galois group of @xmath111 , and @xmath112 the subgroup fixing @xmath0 . then the @xmath45 left cosets of @xmath112 in @xmath110 can be viewed as the embeddings of @xmath0 in @xmath101 , and this makes the cm - type @xmath104 into a set of @xmath29 left cosets of @xmath112 for which we have @xmath113 . let @xmath114 be the union of the left cosets in @xmath104 , and put @xmath115 . let @xmath116 be the stabilizer of @xmath114 in @xmath110 . then @xmath117 defines a subfield @xmath108 of @xmath101 , and as we have @xmath118 we can interpret @xmath119 as a union of left cosets of @xmath117 inside @xmath110 . these cosets define a set of embeddings @xmath109 of @xmath108 into @xmath101 . we call @xmath108 the _ reflex field _ of @xmath120 and we call @xmath109 the _ reflex type_. [ l : reflex ] the field @xmath108 is a cm - field . it is generated over @xmath121 by the sums @xmath122 for @xmath123 , and @xmath109 is a cm - type of @xmath108 . the type norm @xmath107 maps @xmath0 to @xmath108 . the first two statements are proved in ( * ? ? ? * chapter ii , proposition 28 ) ( though the definition of @xmath117 differs from ours , because shimura lets @xmath110 act from the right ) . for the last statement , notice that for @xmath124 , we have @xmath125 , so @xmath126 . a cm - type @xmath104 of @xmath0 is _ induced _ from a cm - subfield @xmath127 if it is of the form @xmath128 for some cm - type @xmath129 of @xmath130 . in other words , @xmath104 is induced from @xmath130 if and only if @xmath114 as above is a union of left cosets of @xmath131 . we call @xmath104 _ primitive _ if it is not induced from a strict subfield of @xmath0 ; primitive cm - types correspond to simple abelian varieties @xcite . notice that the reflex type @xmath109 is primitive by definition of @xmath108 , and that @xmath120 is induced from the reflex of its reflex . in particular , if @xmath104 is primitive , then the reflex of its reflex is @xmath120 itself . for @xmath0 galois and @xmath104 primitive we have @xmath132 , and the reflex type of @xmath104 is @xmath133 . for cm - fields @xmath0 of degree @xmath134 or @xmath135 with primitive cm - types , the reflex field @xmath108 has the same degree as @xmath0 . this fails to be so for @xmath136 . [ l : reflexdeg ] if @xmath0 has degree @xmath45 , then the degree of @xmath108 divides @xmath137 . we have @xmath138 , with @xmath139 totally real and @xmath140 totally negative . the normal closure @xmath101 of @xmath0 is obtained by adjoining to the normal closure @xmath141 of @xmath139 , which has degree dividing @xmath142 , the square roots of the @xmath29 conjugates of @xmath143 . thus @xmath101 is of degree dividing @xmath137 , and @xmath108 is a subfield of @xmath101 . for a ` generic ' cm field @xmath0 the degree of @xmath101 is exactly @xmath144 , and @xmath145 is a field of degree @xmath146 generated by @xmath147 , with @xmath65 ranging over @xmath148 . from and lemma [ l : reflex ] , we find that for every @xmath149 , the element @xmath150 is an element of @xmath151 that satisfies @xmath152 . to make @xmath22 satisfy the conditions of proposition [ p : exist - embed ] , we need to impose conditions modulo @xmath7 on @xmath77 in @xmath108 . suppose @xmath7 splits completely in @xmath0 , and therefore in its normal closure @xmath101 and in the reflex field @xmath108 with respect to @xmath104 . pick a prime @xmath153 over @xmath7 in @xmath101 , and write @xmath154 for @xmath155 . then the factorization of @xmath7 in @xmath156 is @xmath157 [ t : construct - pi ] let @xmath120 be a cm - type and @xmath158 its reflex . let @xmath159 be a prime that splits completely in @xmath0 , and write its factorization in @xmath156 as in . given @xmath160 , write @xmath161 and @xmath162 for @xmath163 . if we have @xmath164 for some primitive @xmath8-th root of unity @xmath165 , then @xmath166 satisfies @xmath152 and @xmath167 this is a straightforward generalization of the argument in example [ cycliccase ] . the conditions generalize and , and imply in the present context that @xmath168 and @xmath169 are in the prime @xmath170 over @xmath7 that underlies the factorization . if the element @xmath22 in theorem [ t : construct - pi ] generates @xmath0 and @xmath171 is a prime @xmath3 that is unramified in @xmath0 , then by lemma [ l : honda - tate ] @xmath22 is a @xmath3-weil number corresponding to an ordinary abelian variety @xmath5 over @xmath172 with endomorphism algebra @xmath0 and frobenius element @xmath22 . by proposition [ p : exist - embed ] , @xmath5 has embedding degree @xmath8 with respect to @xmath7 . this leads to the following algorithm . [ alg : construct - pi ] input : a cm - field @xmath0 of degree @xmath173 , a primitive cm - type @xmath104 of @xmath0 , a positive integer @xmath8 , and a prime @xmath159 that splits completely in @xmath0 . output : a prime @xmath3 and a @xmath3-weil number @xmath174 corresponding to an ordinary , simple abelian variety @xmath175 with embedding degree @xmath8 with respect to @xmath7 . 1 . compute a galois closure @xmath101 of @xmath0 and the reflex @xmath158 of @xmath120 . set @xmath176 and write @xmath177 . [ st : factor ] fix a prime @xmath178 of @xmath179 , and compute the factorization of @xmath7 in @xmath156 as in . compute a primitive @xmath8-th root of unity @xmath165 . [ st : random ] choose random @xmath180 . 5 . set @xmath181 and @xmath182 . 6 . [ st : crt ] compute @xmath183 such that @xmath184 and @xmath185 for @xmath186 . [ st : q ] set @xmath187 . if @xmath3 is not prime , go to step . [ st : pi ] set @xmath188 . if @xmath3 is not unramified in @xmath0 , or @xmath22 does not generate @xmath0 , go to step . [ st : output ] return @xmath3 and @xmath22 . we require @xmath189 in algorithm [ alg : construct - pi ] , as the case @xmath35 is already covered by example [ cycliccase ] , and requires a slight adaptation . the condition that @xmath7 be prime is for simplicity of presentation only ; the algorithm easily extends to square - free values of @xmath7 that are given as products of splitting primes . such @xmath7 are required , for example , by the cryptosystem of @xcite . [ thm : ispoltime ] if the field @xmath0 is fixed , then the heuristic expected run time of algorithm [ alg : construct - pi ] is polynomial in @xmath28 . the algorithm consists of a precomputation for the field @xmath0 in steps ( 1)(3 ) , followed by a loop in steps that is performed until an element @xmath77 is found that has prime norm @xmath190 , and we also find in step that @xmath3 is unramified in @xmath0 and the type norm @xmath191 generates @xmath0 . the primality condition in step is the ` true ' condition that becomes harder to achieve with increasing @xmath7 , whereas the conditions in step , which are necessary to guarantee correctness of the output , are so extremely likely to be fulfilled ( especially in cryptographic applications where @xmath0 is small and @xmath7 is large ) that they will hardly ever fail in practice and only influence the run time by a constant factor . as @xmath77 is computed in step as the lift to @xmath156 of an element @xmath192 , its norm can be bounded by a constant multiple of @xmath193 . heuristically , @xmath194 behaves as a random number , so by the prime number theorem it will be prime with probability at least @xmath195 , and we expect that we need to repeat the loop in steps about @xmath196 times before finding @xmath77 of prime norm @xmath3 . as each of the steps is polynomial in @xmath28 , so is the expected run time up to step , and we are done if we show that the conditions in step are met with some positive probability if @xmath0 is fixed and @xmath7 is sufficiently large . for @xmath3 being unramified in @xmath0 , one simply notes that only finitely many primes ramify in the field @xmath0 ( which is fixed ) and that @xmath3 tends to infinity with @xmath7 , since @xmath7 divides @xmath197 . finally , we show that @xmath22 generates @xmath0 with probability tending to @xmath33 as @xmath7 tends to infinity . suppose that for every vector @xmath198 that is not all 0 or 1 , we have @xmath199 this set of @xmath200 ( dependent ) conditions on the @xmath201 independent random variables @xmath202 for @xmath203 is satisfied with probability at least @xmath204 . for any automorphism @xmath205 of @xmath101 , the set @xmath206 is a cm - type of @xmath145 and there is a @xmath198 such that @xmath207 if @xmath208 contains @xmath209 and @xmath210 otherwise . then @xmath80 is @xmath211 , while @xmath212 is @xmath213 , so @xmath214 is @xmath215 . by , if this expression is @xmath33 then @xmath216 or @xmath217 , so @xmath218 or @xmath219 , which by definition of the reflex is equivalent to @xmath205 or @xmath220 being trivial on @xmath0 , i.e. , to @xmath205 being trivial on the maximal real subfield @xmath139 . thus if holds , then @xmath221 implies that @xmath205 is trivial on @xmath139 , hence @xmath222 . since @xmath174 is not real ( otherwise , @xmath223 ramifies in @xmath0 ) , this implies that @xmath224 . in order to maximize the likelihood of finding prime norms , one should minimize the norm of the lift @xmath77 computed in the chinese remainder step . this involves minimizing a norm function of degree @xmath225 in @xmath225 integral variables , which is already infeasible for @xmath226 . in practice , for given @xmath7 , one lifts a standard basis of @xmath227 to @xmath156 . multiplying those lifts by integer representatives for the elements @xmath80 and @xmath212 of @xmath228 , one quickly obtains lifts @xmath77 . we also choose , independently of @xmath7 , a @xmath229-basis of @xmath230 consisting of elements that are ` small ' with respect to all absolute values of @xmath145 . we translate @xmath77 by multiples of @xmath7 to lie in @xmath231 , where @xmath232 is the fundamental parallelotope in @xmath233 consisting of those elements that have coordinates in @xmath234 $ ] with respect to our chosen basis . if we denote the maximum on @xmath235 of all complex absolute values of @xmath108 by @xmath236 , we have @xmath237 for the @xmath32-value we find @xmath238 which is approximately @xmath239 if @xmath7 gets large with respect to @xmath236 . we would like @xmath32 to be small , but this is not what one obtains by lifting random admissible choices of @xmath240 . [ thm : heur ] if the field @xmath0 is fixed and @xmath7 is large , we expect that ( 1 ) the output @xmath3 of algorithm [ alg : construct - pi ] yields @xmath241 , and ( 2 ) an optimal choice of @xmath242 satisfying the conditions of theorem [ t : construct - pi ] yields @xmath243 . find an efficient algorithm to compute an element @xmath242 satisfying the conditions of theorem [ t : construct - pi ] for which @xmath243 . we will prove theorem [ thm : heur ] via a series of lemmas . let @xmath244 be the subset of the parallelotope @xmath245 consisting of those @xmath246 that satisfy the two congruence conditions for a given embedding degree @xmath8 . heuristically , we will treat the elements of @xmath244 as random elements of @xmath231 with respect to the distributions of complex absolute values and norm functions . we will also use the fact that , as @xmath108 is totally complex of degree @xmath247 , the @xmath248-algebra @xmath249 is naturally isomorphic to @xmath250 . we assume throughout that @xmath251 . [ l : lower - max ] fix the field @xmath0 . under our heuristic assumption , there exists a constant @xmath252 such that for all @xmath253 , the probability that a random @xmath254 satisfies @xmath255 is less than @xmath256 . the probability that a random @xmath77 lies in the set @xmath257 is the quotient of the volume of @xmath258 by the volume @xmath259 of @xmath231 , where @xmath260 is the discriminant of @xmath108 . now @xmath258 is contained inside @xmath261 , which has volume @xmath262^{{\widehat{g}}}\\ \prod { \lvert x_i \rvert}^2\leq r^{2({{\widehat{g}}}-\varepsilon ) } } $ } } \prod\|x_i\| dx \ \ < \ \ ( 2\pi)^{{\widehat{g}}}\mathop{\int}_{\hbox to0pt{\hss $ \scriptstyle x\in [ 0,rm_{\widehat{k}}]^{{\widehat{g}}}$ } } r^{{{\widehat{g}}}-\varepsilon}dx \ \ = \ \ ( 2\pi m_{\widehat{k}})^{{\widehat{g}}}r^{2{{\widehat{g}}}-\varepsilon},\ ] ] so a random @xmath77 lies in @xmath258 with probability less than @xmath263 . [ lem : aux ] there exists a number @xmath264 , depending only on @xmath145 , such that for any positive real number @xmath265 , the expected number of @xmath254 with all absolute values below @xmath266 is @xmath267 let @xmath268 be a lower bound on @xmath269 for the maximum of all complex absolute values , so the box @xmath270 consisting of those elements that have all absolute values below @xmath266 lies completely inside @xmath271 . the volume of @xmath272 in @xmath249 is @xmath273 , while @xmath231 has volume @xmath274 . the expected number of @xmath254 satisfying @xmath275 for all absolute values is @xmath276 times the quotient of these volumes . [ l : upper - min ] fix the field @xmath0 . under our heuristic assumption , there exists a constant @xmath277 such that for all positive @xmath278 , if @xmath7 is sufficiently large , then we expect the number of @xmath254 satisfying @xmath279 to be at least @xmath280 . any @xmath77 as in lemma [ lem : aux ] satisfies @xmath281 , so we apply the lemma to @xmath282 , which is less than @xmath283 for large enough @xmath7 and @xmath284 . [ l : lower - min ] fix the field @xmath0 . under our heuristic assumption , for all @xmath253 , if @xmath7 is large enough , we expect there to be no @xmath254 satisfying @xmath285 . let @xmath286 be the ring of integers of the maximal real subfield of @xmath145 . let @xmath287 be the subgroup of norm one elements of @xmath288 . we embed @xmath287 into @xmath289 by mapping @xmath290 to the vector @xmath291 of logarithms of absolute values of @xmath292 . the image is a complete lattice in the @xmath293-dimensional space of vectors with coordinate sum @xmath294 . fix a fundamental parallelotope @xmath295 for this lattice . let @xmath296 be the element of @xmath244 of smallest norm . since the conditions , as well as the norm of @xmath296 , are invariant under multiplication by elements of @xmath287 , we may assume without loss of generality that @xmath297 is inside @xmath298 . then every difference of two entries of @xmath297 is bounded , and hence every quotient of absolute values of @xmath296 is bounded from below by a positive constant @xmath299 depending only on @xmath0 . in particular , if @xmath300 is the maximum of all absolute values of @xmath296 , then @xmath301 . now suppose @xmath296 has norm below @xmath302 . then all absolute values of @xmath296 are below @xmath303 , and @xmath304 for @xmath7 sufficiently large . now lemma [ lem : aux ] implies that the expected number of @xmath305 with all absolute values below @xmath266 is a constant times @xmath306 , so for any sufficiently large @xmath7 we expect there to be no such @xmath77 , a contradiction . the upper bound @xmath307 follows from . lemma [ l : lower - max ] shows that for any @xmath253 , the probability that @xmath32 is smaller than @xmath308 tends to zero as @xmath7 tends to infinity , thus proving the lower bound @xmath309 . lemma [ l : upper - min ] shows that for any @xmath253 , if @xmath7 is sufficiently large then we expect there to exist a @xmath77 with @xmath32-value at most @xmath310 , thus proving the bound @xmath311 . lemma [ l : lower - min ] shows that we expect @xmath312 for the optimal @xmath77 , which proves the bound @xmath313 . for very small values of @xmath7 we are able to do a brute - force search for the smallest @xmath3 by testing all possible values of @xmath314 in step [ st : random ] of algorithm [ alg : construct - pi ] . we performed two such searches , one in dimension 2 and one in dimension 3 . the experimental results support our heuristic evidence that @xmath243 is possible with a smart choice in the algorithm , and that @xmath315 is achieved with a randomized algorithm . take @xmath316 , and let @xmath317 be the cm - type of @xmath0 defined by @xmath318 . we ran algorithm [ alg : construct - pi ] with @xmath319 and @xmath320 , and tested all possible values of @xmath321 . the total number of primes @xmath3 found was @xmath322 , and the corresponding @xmath32-values were distributed as follows : @xmath323{plot1}\quad\includegraphics[width=4.5cm]{plot2}\ ] ] the smallest @xmath3 found was @xmath324 , giving a @xmath32-value of @xmath325 . the curve over @xmath326 for which the jacobian has this @xmath32-value is @xmath327 , and the number of points on its jacobian is @xmath328 . take @xmath329 , and let @xmath330 be the cm - type of @xmath0 defined by @xmath331 . we ran algorithm [ alg : construct - pi ] with @xmath332 and @xmath333 , and tested all possible values of @xmath334 . the total number of primes @xmath3 found was @xmath335 , and the corresponding @xmath32-values were distributed as follows : @xmath323{plot3}\quad\includegraphics[width=4.5cm]{plot4}\ ] ] the smallest @xmath3 found was @xmath336 , giving a @xmath32-value of @xmath337 . the curve over @xmath326 for which the jacobian has this @xmath32-value is @xmath338 , and the number of points on its jacobian is @xmath339 . take @xmath316 , and let @xmath317 be the cm - type of @xmath0 defined by @xmath340 . we ran algorithm [ alg : construct - pi ] with @xmath341 and @xmath342 , and tested @xmath343 random values of @xmath321 . the total number of primes @xmath3 found was @xmath344 . of these primes , 6509 ( 91.6% ) produced @xmath32-values between 7.9 and 8.0 , while 592 ( 8.3% ) had @xmath32-values between 7.8 and 7.9 . the smallest @xmath3 found had @xmath345 binary digits , giving a @xmath32-value of @xmath346 . our algorithm [ alg : construct - pi ] yields @xmath3-weil numbers @xmath174 that correspond , in the sense of honda and tate @xcite , to isogeny classes of ordinary , simple abelian varieties over prime fields that have a point of order @xmath7 and embedding degree @xmath8 with respect to @xmath7 . it does not give a method to explicitly construct an abelian variety @xmath5 with frobenius @xmath174 . in this section we focus on the problem of explicitly constructing such varieties using complex multiplication techniques . the key point of the complex multiplication construction is the fact that every ordinary , simple abelian variety over @xmath172 with frobenius @xmath174 arises as the reduction at a prime over @xmath3 of some abelian variety @xmath347 in characteristic zero that has cm by the ring of integers of @xmath0 . thus if we have fixed our @xmath0 as in algorithm [ alg : construct - pi ] , we can solve the construction problem for all ordinary weil numbers coming out of the algorithm by compiling the finite list of @xmath348-isogeny classes of abelian varieties in characteristic zero having cm by @xmath34 . there will be one @xmath348-isogeny class for each equivalence class of primitive cm - types of @xmath0 , where @xmath104 and @xmath129 are said to be equivalent if we have @xmath349 for an automorphism @xmath65 of @xmath0 . as we can choose our favorite field @xmath0 of degree @xmath45 to produce abelian varieties of dimension @xmath29 , we can pick fields @xmath0 for which such lists already occur in the literature . from representatives of our list of isogeny classes of abelian varieties in characteristic zero having cm by @xmath34 , we obtain a list @xmath350 of abelian varieties over @xmath6 with cm by @xmath34 by reducing at some fixed prime @xmath351 over @xmath3 . changing the choice of the prime @xmath351 amounts to taking the reduction at @xmath351 of a conjugate abelian variety , which also has cm by @xmath34 and hence is @xmath352-isogenous to one already in the list . for every abelian variety @xmath353 , we compute the set of its twists , i.e. , all the varieties up to @xmath6-isomorphism that become isomorphic to @xmath5 over @xmath352 . there is at least one twist @xmath354 of an element @xmath355 satisfying @xmath356 , and this @xmath354 has a point of order @xmath7 and the desired embedding degree . note that while efficient point - counting algorithms do not exist for varieties of dimension @xmath357 , we can determine probabilistically whether an abelian variety has a given order by choosing a random point , multiplying by the expected order , and seeing if the result is the identity . the complexity of the construction problem rapidly increases with the genus @xmath358/2 $ ] , and it is fair to say that we only have satisfactory general methods at our disposal in very small genus . in genus one , we are dealing with elliptic curves . the @xmath359-invariants of elliptic curves over @xmath23 with cm by @xmath34 are the roots of the _ hilbert class polynomial _ of @xmath0 , which lies in @xmath360 $ ] . the degree of this polynomial is the class number @xmath361 of @xmath0 , and it can be computed in time @xmath362 . for genus 2 , we have to construct abelian surfaces . any principally polarized abelian surface is the jacobian of a genus 2 curve , and all genus 2 curves are hyperelliptic . there is a theory of class polynomials analogous to that for elliptic curves , as well as several algorithms to compute these polynomials , which lie in @xmath363 $ ] . the genus 2 algorithms are not as well - developed as those for elliptic curves ; at present they can handle only very small quartic cm - fields , and there exists no rigorous run time estimate . from the roots in @xmath6 of these polynomials , we can compute the genus 2 curves using mestre s algorithm . any three - dimensional principally polarized abelian variety is isogenous to the jacobian of a genus 3 curve . there are two known families of genus 3 curves over @xmath23 whose jacobians have cm by an order of dimension @xmath364 . the first family , due to weng @xcite , gives hyperelliptic curves whose jacobians have cm by a degree-6 field containing @xmath365 . the second family , due to koike and weng @xcite , gives picard curves ( curves of the form @xmath366 with @xmath367 ) whose jacobians have cm by a degree-6 field containing @xmath368 . explicit cm - theory is mostly undeveloped for dimension @xmath369 . moreover , most principally polarized abelian varieties of dimension @xmath370 are not jacobians , as the moduli space of jacobians has dimension @xmath371 , while the moduli space of abelian varieties has dimension @xmath372 . for implementation purposes we prefer jacobians or even hyperelliptic jacobians , as these are the only abelian varieties for which group operations can be computed efficiently . in cases where we can not compute every abelian variety in characteristic zero with cm by @xmath34 , we use a single such variety @xmath5 and run algorithm [ alg : construct - pi ] for each different cm - type of @xmath0 until it yields a prime @xmath3 for which the reduction of @xmath5 mod @xmath3 is in the correct isogeny class . an example for @xmath373 with @xmath27 prime is given by the jacobian of @xmath374 , which has dimension @xmath375 . we implemented algorithm [ alg : construct - pi ] in magma and used it to compute examples of hyperelliptic curves of genus 2 and 3 over fields of cryptographic size for which the jacobians are pairing - friendly . the subgroup size @xmath7 is chosen so that the discrete logarithm problem in @xmath14 $ ] is expected to take roughly @xmath376 steps . the embedding degree @xmath8 is chosen so that @xmath377 ; this would be the ideal embedding degree for the 80-bit security level if we could construct varieties over @xmath378 with @xmath379 . space constraints prevent us from giving the group orders for each jacobian , but we note that a set of all possible @xmath3-weil numbers in @xmath0 , and hence all possible group orders , can be computed from the factorization of @xmath3 in @xmath0 . let @xmath380 and let @xmath0 be the degree-4 galois cm field @xmath381 . let @xmath317 be the cm type of @xmath0 such that @xmath382 . we ran algorithm [ alg : construct - pi ] with cm type @xmath120 , @xmath383 , and @xmath384 . the algorithm output the following field size : @xmath385 there is a single @xmath386-isomorphism class of curves over @xmath387 whose jacobians have cm by @xmath34 and it has been computed in @xcite ; the desired twist turns out to be @xmath388 . the @xmath32-value of @xmath389 is @xmath390 . let @xmath391 and let @xmath0 be the degree-4 non - galois cm field @xmath381 . the reflex field @xmath108 is @xmath392 where @xmath393 . let @xmath109 be the cm type of @xmath0 such that @xmath394 . we ran algorithm [ alg : construct - pi ] with the cm type @xmath395 , subgroup size @xmath396 , and embedding degree @xmath384 . the algorithm output the following field size : @xmath397 the class polynomials for @xmath0 can be found in the preprint version of @xcite . we used the roots of the class polynomials mod @xmath3 to construct curves over @xmath387 with cm by @xmath34 . as @xmath0 is non - galois with class number @xmath135 , there are 8 isomorphism classes of curves in 2 isogeny classes . we found a curve @xmath398 in the correct isogeny class with equation @xmath399 , with @xmath400 the @xmath32-value of @xmath389 is @xmath401 . let @xmath0 be the degree-6 galois cm field @xmath402 , and let @xmath330 be the cm type of @xmath0 such that @xmath403 . we used the cm type @xmath120 to construct a curve @xmath398 whose jacobian has embedding degree @xmath404 with respect to @xmath405 . since @xmath0 has class number @xmath33 and one equivalence class of primitive cm types , there is a unique isomorphism class of curves in characteristic zero whose jacobians are simple and have cm by @xmath0 ; these curves are given by @xmath406 . algorithm [ alg : construct - pi ] output the following field size : @xmath407 the equation of the curve @xmath398 is @xmath408 . the @xmath32-value of @xmath409 is @xmath410 . we conclude with an example of an 8-dimensional abelian variety found using our algorithms . we started with a single cm abelian variety @xmath5 in characteristic zero and applied our algorithm to different cm - types until we found a prime @xmath3 for which the reduction has the given embedding degree . let @xmath411 . we set @xmath319 and @xmath412 and ran algorithm [ alg : construct - pi ] repeatedly with different cm types for @xmath0 . given the output , we tested the jacobians of twists of @xmath413 for the specified number of points . we found that the curve @xmath414 has embedding degree @xmath415 with respect to @xmath7 over the field @xmath6 of order @xmath416 the cm type was @xmath417 where @xmath418 . the @xmath32-value of @xmath389 is @xmath419 .	we present an algorithm that , on input of a cm - field @xmath0 , an integer @xmath1 , and a prime @xmath2 , constructs a @xmath3-weil number @xmath4 corresponding to an ordinary , simple abelian variety @xmath5 over the field @xmath6 of @xmath3 elements that has an @xmath6-rational point of order @xmath7 and embedding degree @xmath8 with respect to @xmath7 . we then discuss how cm - methods over @xmath0 can be used to explicitly construct @xmath5 .
gamma - ray emitting pulsar wind nebulae ( pwne ) are excellent test grounds for studying pulsars relativistic winds and particle acceleration that takes place presumably at termination shocks . especially , pwne which accompany the diffuse synchrotron x - ray emission show the important clues to understand the time evolution of the pwne and how the accelerated particles escape from the shocks . while young pwne with characteristic ages of less than 10 kyr are relatively well studied in the past observations with _ chandra _ and _ xmm_-_newton _ @xcite , the sample of middle aged pwne remains small . the kookaburra region includes a middle aged pwn ( psr j1420 - 6048 ) and a plausible pwn ( rabbit ) , making it a suitable target for the detailed study of the diffuse emission with the _ suzaku _ x - ray observatory . _ suzaku _ , characterized by the low detector background compared to _ chandra _ and _ xmm - newton _ , is crucial for the analysis of faint and diffuse x - ray emission . the complex of compact and extended radio / x - ray sources , called kookaburra ( designated by @xcite ) , spans over about one square degree along the galactic plane around @xmath8 . the h.e.s.s . galactic survey revealed two very high energy ( vhe ) sources in this region ; the brighter of the two sources , hess j1420 - 607 , is centered at the position of ( r.a . = ( 14@xmath920@xmath1009@xmath11 , -60@xmath1245@xmath1336@xmath14 ) with an intrinsic extension of @xmath15 , and the slightly less bright second source , hess j1418 - 609 , is centered at the position of ( r.a . , = ( 14@xmath918@xmath1004@xmath11 , -60@xmath1258@xmath1331@xmath14 ) with an intrinsic extension of @xmath16 ( a major - axis of @xmath17 and a minor - axis of @xmath18 fitted with an elongated gaussian shape ) . the kookaburra region is also bright in the gev band and registered as a _ @xmath0-ray source @xcite . there is an energetic pulsar psr j1420 - 6048 at the south of hess j1420 - 607 . it is located at ( r.a . , dec . ) = ( 14@xmath920@xmath1008@xmath19 , -60@xmath1248@xmath1317@xmath20 ) , which is @xmath21 offset from the center of the vhe @xmath0-ray emission . psr j1420 - 6048 is a radio / x - ray pulsar with a period of @xmath22 ms and period derivative of @xmath23 @xcite . the distance of the pulsar was estimated as @xmath24 kpc based on the pulsar s dispersion measure with the ne2001 galactic electron - density model @xcite . the characteristic age and the spin - down luminosity are @xmath25 kyr and @xmath26 erg s@xmath27 , respectively . the strength of surface magnetic field is @xmath28 g. _ chandra _ observations revealed a faint and narrow diffuse x - ray emission in k3 nebula ( designated by @xcite with the radio observation ) around psr j1420 - 6048 @xcite . spectral fitting with an absorbed power - law model to the k3 nebula ( @xmath29 radius from the pulsar ) yielded an absorption column density of @xmath30 with a photon index of @xmath31 . the rabbit nebula is located at the eastern edge of hess j1418 - 609 with a distance of @xmath32 to the best central position of the vhe emission . @xcite found two point - like sources with the _ xmm - newton _ data , labeled as r1 and r2 ; the brighter source ( r1 ) is located at the edge of the rabbit , while the fainter source ( r2 ) at ( r.a . = ( 14@xmath918@xmath1039@xmath33 , -60@xmath1257@xmath1356@xmath34 ) , which appears embedded in the narrow diffuse emission . timing analysis with the epic pn data revealed a @xmath35 ms pulsation with a period derivative of @xmath36 from r2 , although not highly significant . assuming these values are correct , the characteristic age can be estimated as @xmath37 kyr . fitting the spectrum from the whole rabbit nebula ( @xmath38 radius ) with an absorbed power - law model , they found an absorption column density of @xmath39 and a photon index of @xmath40 . the two vhe sources are considered to be pwne @xcite , given the x - ray and tev results reported so far . generally , detailed observations of the synchrotron x - ray nebulae are crucial to understand spatial and spectral distributions of accelerated electrons . however , previous x - ray observations did not have enough sensitivity to detect faint x - ray nebulae with sizes comparable to the tev nebulae . in this paper , we report new detections of the diffuse x - ray emission of the kookaburra complex using the _ suzaku _ satellite . we observed the kookaburra region with the _ suzaku _ satellite in 2009 february . the _ suzaku _ observations were performed with the x - ray imaging spectrometer ( xis : @xcite ) in 0.3 12 kev and the hard x - ray detector ( hxd : @xcite ) in 13 600 kev . the xis , located at the focal plane of the x - ray telescopes ( xrt : @xcite ) , consists of one back - illuminated ccd camera ( xis1 ) and three front - illuminated ccds ( xis0 , 2 , and 3 ) . one of the front - illuminated ccds , xis2 , was not available at the time of our observations , since it suffered from a fatal damage on 9 november 2006 , and unusable since then . the xis instruments were operated in a normal full - frame clocking mode ( a frame time of 8 s ) with spaced - row charge injection ( sci ) @xcite . the hxd consists of the silicon pin photo diodes ( hereafter pin ) capable of observations in the 13 70 kev band and the gso crystal scintillators ( hereafter gso ) which cover the 40 600 kev band . since we could not find significant signals in the hxd data after subtracting the non - x - ray background ( nxb ) , cosmic x - ray background ( cxb ) , and galactic ridge components , we focused on the xis data analysis in this paper . we used data sets processed by a set of software of the _ suzaku _ data processing pipeline ( version 2.2.11.24 ) . the telemetry saturating time was excluded in the pipeline processing . basic analysis was done using the heasoft software package ( version 6.4 ) . we made use of cleaned event files , in which standard screening was applied . the standard screening procedures include event grade selections , and removal of time periods such as spacecraft passage of the south atlantic anomaly ( saa ) , intervals of small geomagnetic cutoff rigidity ( cor ) , and those of a low elevation angle . specifically , for the xis , the elevation angle larger than 5@xmath12 above the earth and larger than 20@xmath12 from the sunlit earth limb are selected . the total exposure from the two observations amounts to 72 ks after standard data screening . table [ tab : kookaburra_obs ] gives the log of the _ suzaku _ observations . i d & coord . ( j2000 ) & exposure & date + & r.a . , dec . & xis & + 503110010 & 215.0292 , -60.8156 & 50.3 & 01/11/2009 + 503071010 & 214.6625 , -60.9675 & 21.3 & 02/14/2009 + [ tab : kookaburra_obs ] in order to combine two different pointing images , we extracted the photon events in the energy range of 210 kev from each sensor . the data between 5.73 and 6.67 kev were removed from the image to exclude the calibration sources . we corrected the vignetting effect by dividing the image with a flat sky image simulated in the energy range of 28 kev using the xrt + xis simulator @xcite . the image was binned to 8 @xmath41 8 pixels and smoothed with a gaussian function of @xmath42 . a combined _ suzaku _ xis ( 0 + 1 + 3 ) image in the kookaburra region is show in fig . [ fig : image_merge ] . * both hess j1420 - 607 and hess j1418 - 609 are reported as diffuse tev-@xmath0-ray emitting sources @xcite . the intrinsic extensions of the diffuse emissions are indicated as white circles in the figure . the overlaid contours are taken from the h.e.s.s . observation . we can clearly see that the separate x - ray sources locate with some offsets from the @xmath0-ray peak positions and are coincident with the vhe sources . * we extracted the photon events in the energy ranges of 12 kev and 210 kev . the xis images are shown in fig . [ fig : k3image ] . we can see several thermal sources in the lower energy image , which are invisible in the higher energies . the bright x - ray emission is well coincident with the location of psr j1420 - 6048 and extended diffuse emission surrounds the bright pulsar . the diffuse component has an elongated shape which extends from the pulsar to the @xmath0-ray peak . * in order to determine the extension of non - thermal x - ray emission , we created a surface brightness profile from the enclosed region shown as `` rect '' in fig . [ fig : k3image ] ( a ) along the north to south direction . the surface brightness in 210 kev vs. relative coordinate is shown in fig . [ fig:1d_k3pos ] . the profile was fitted with a gaussian function plus constant as a background to derive the size of the diffuse emission , using the sigma of the gaussian ( @xmath43 ) . since a gaussian profile often provides a reasonable approximation for the surface brightness profiles @xcite , we used relative coordinates between [email protected] and [email protected] as a fitting range in consideration for the non - symmetrical profile . we excluded a range between [email protected] and 4@xmath44 as a narrow pulsar component . this is to avoid the contribution from the bright pulsar , whose image is smeared by the point - spread function of xrt . we defined the source size as three times the @xmath43 . the fitted result shows @xmath45 . if we assume the distance as 5.6 kpc , the physical size of the extended diffuse emission corresponds to @xmath46 pc . the intrinsic emission size of the bright pulsar was also estimated by fitting the image with a two - dimensional gaussian function , however , the calculated size was smaller than the attitude fluctuation of the satellite ( see * ? ? ? * ) . * ( a ) . ] fig . [ fig : raimage ] shows the xis images of the rabbit nebula . similar to k3/psr j1420 - 6048 , thermal sources are visible in the lower energies . the x - ray peak is coincident with the point - like sources reported by @xcite , which are indicated as black cross points in fig . [ fig : raimage ] . according to @xcite , these bright sources are embedded in narrow diffuse emission . compared with the _ xmm_-_newton _ , the diffuse emission in fig . [ fig : raimage ] ( b ) looks more extended than the previous image , and another diffuse structure which is clearly separated from the point - like sources can be seen in the southeast direction . in order to determine the size of the diffuse emission , we made a projected profile with the same procedure as in k3/psr j1420 - 6048 . the extracted area is indicated as `` rect '' in fig . [ fig : raimage ] ( a ) . the 1d projected result is shown in fig . [ fig:1d_rapos ] . we excluded a range between @xmath47 and @xmath48 from the fitting as the central pulsar region to avoid its influence . the projected profile was well represented with a gaussian function plus constant background component . the size of the diffuse component is estimated as @xmath49 from the 1d profile . the physical size of the diffuse emission can be estimated as @xmath50 pc . we assumed the distance as 5 kpc . ( a ) . ] firstly , we checked the background level extracted from the `` bgd '' region in fig . [ fig : k3image ] ( b ) . we compared its background spectrum with public blank sky observations on lockman - hole ( obs . i d : 102018010 ) and also with aso 0304 ( obs . i d : 504054010 ) , which locates near the kookaburra region , i.e. ( r.a . , dec.)=(213.3355 , -62.0808 ) . the background levels are consistent within 10% and the galactic ridge emission is ignorable . we thus concluded that the `` bgd '' region is a source - free in the following spectral analysis . in order to determine the absorption column density around psr j1420 - 6048 , we extracted the photon events within an angular distance of 1@xmath51 from the pulsar , which corresponds to `` ring1+ring2 '' in fig . [ fig : k3image ] ( b ) . we have co - added the data from xis0 and xis3 to increase statistics . the response ( rmf ) files and the auxiliary response ( arf ) files were produced using _ xisrmfgen _ and _ xissimarfgen _ , respectively . [ fig : spec_k3_src ] shows the spectral result with the joint - fitted xis0 + 3 ( black ) and xis1 ( red ) data . the spectral shapes are well represented by an absorbed power - law model ( ` wabs@xmath41pow ' ) . from the spectral fitting , the absorption column density was estimated as @xmath52 , which is consistent with the previous observations . to determine the total flux from the pulsar and diffuse emission , we used the `` reg1-reg2 '' as the extracted region . the flux in the energy range of 210 kev is @xmath53 erg@xmath54 . the systematic uncertainty coming from the difference of the effective area between the arf responses for diffuse and point sources was less than 10% . in order to investigate the spatial dependence of the spectral shapes , we separated the extraction region into four annular regions , i.e. `` ring1 '' : 0@xmath55 , `` ring2 '' : @xmath55@xmath56 , `` ring3 '' : @xmath56@xmath57 , `` ring4 '' : @xmath57@xmath58 . to reduce the uncertainty of the column density with limited statistics , we fixed the absorption column density at @xmath59 determined from the `` ring1+ring2 '' region . [ fig : k3_4specs ] shows the spectra ( xis0 + 3 ) for each circular region . in order to exclude the thermal contamination , we used the 2 - 10 kev band for the spectral fitting . in the `` ring3 '' spectrum , we can see a weak trend of the iron emission line around 6.7 kev . this might come from unresolved thermal sources . the photon index vs. relative coordinate is shown in fig . [ fig : k3_index_change ] . quoted errors are at the @xmath60 confidence level . we can see that the photon index slightly changes according to the distance from the bright pulsar region . the contamination effect from a point source in `` ring1 '' on the outer region `` ring4 '' was estimated as @xmath615% . the fitted parameters are summarized in table [ tab : fit ] . as for the hxd - pin , we could not detect any significant pulse in timing analysis , thus there is no excess in the hard x - ray band from the pulsar . ( b ) . ] we performed the same analytical procedures to the photon events from the rabbit nebula as described in k3/psr j1420 - 6048 . [ fig : spec_ra_src ] shows the spectra extracted within an angular distance of @xmath62 from the x - ray peak , which corresponds to the `` ring1+ring2 '' region in fig . [ fig : raimage ] ( b ) . the background spectrum was consistent with that of the blank sky data . the spectral shapes are well represented by an absorbed power - law model . the absorption column density was determined as @xmath63 by the spectral fitting . for the spatial dependence of the spectral shapes , we chose the annular regions as `` ring1 '' : 0@xmath64 , `` ring2 '' : @xmath64@xmath62 , `` ring3 '' : @xmath62@xmath65 , `` ring4 '' : @xmath65@xmath66 . the spectral results are shown in fig . [ fig : rabbit_spec_4circulars ] . to exclude the thermal contamination , we used the energy range of 210 kev for spectral fittings . the photon index vs. relative coordinate is shown in fig . [ fig : rabbit_index_change ] . the spectral shape becomes softer according to the distance from the inner bright core to the outer region , which is a similar trend as in the k3//psr j1420 - 6048 region . the systematic error of the psf in `` ring1 '' to `` ring4 '' was less than 5% . the best - fit parameters are summarized in table [ tab : fit ] . the total flux in the energy range of 210 kev is estimated as @xmath67 erg@xmath54 determined from the extraction region of `` reg1-reg2 '' . ( b ) . ] rrrrr object & region & @xmath68 ( wabs ) & photon index & flux [ 2 - 10 kev ] & @xmath69 + k3/psr j1420 - 6048 , ring1+ring2 & 4.4 @xmath70 0.3 & 2.00 @xmath70 0.10 & 1.53 @xmath70 0.03 & 1.01 ( 72 ) + reg1-reg2 & 4.4 ( fix ) & @xmath71 & @xmath72 & 0.83 ( 33 ) + + ring1 & & @xmath73 & & 0.86 ( 32 ) + ring2 & & @xmath74 & & 1.01 ( 32 ) + ring3 & &@xmath75 & & 0.86 ( 35 ) + ring4 & &@xmath76 & & 0.94 ( 35 ) + rabbit , ring1+ring2 & 2.7 @xmath70 0.2 & 1.82 @xmath70 0.10 & 2.65 @xmath70 0.05 & 0.87 ( 101 ) + reg1-reg2 & 2.7 ( fix ) & 2.00 @xmath70 0.06 & 6.27@xmath70 0.13 & 0.80 ( 52 ) + + ring1 & & @xmath77 & & 1.06 ( 46 ) + ring2 & & @xmath78 & & 1.09 ( 35 ) + ring3 & & @xmath79 & & 1.10 ( 46 ) + ring4 & & @xmath80 & & 1.08 ( 33 ) + [ tab : fit ] we have detected the diffuse x - ray emission around k3/psr j1420 - 6048 and rabbit , which were unnoticed in the previous observations , in the kookaburra region with the _ suzaku _ satellite . the x - ray peaks of the two sources are both within the error circles of the tev sources and each of them has an offset from the @xmath0-ray peak with @xmath81 for k3/psr j1420 - 6048 and @xmath82 for rabbit . the xis spectra of the two sources were well reproduced by an absorbed power - law model with a photon index of @xmath83 2.3 . if tev / x emission comes from the same object , the origin of the vhe @xmath0-rays can be explained by the inverse compton ( ic ) scattering of the cmb photons by high - energy electrons , while the x - ray emission via synchrotron radiation in a mean magnetic field @xmath84 . in this case , typical energies of responsible electrons are @xmath85 tev for ic @xmath0-rays ( in the thomson regime ) at a photon energy @xmath86 , and @xmath87 tev for x - rays at @xmath88 , respectively . the synchrotron and ic spectra produced by the relativistic electrons obeying an energy distribution of @xmath89 , have the same spectral shape , @xmath90 as long as ic scattering occurs in the thomson regime . if the high - energy electrons are continuously injected into the radiation zone at a constant rate and loosing energy predominantly by synchrotron or ic ( thomson ) losses , the electron energy spectrum becomes one power of @xmath91 steeper above the cooling break @xmath92 that is determined by the cooling timescale and the age of the source . as a result , the synchrotron and ic spectra from the cooled electrons become softer by @xmath93 . our _ suzaku _ results clearly show that the photon index smoothly changes according to the distance from the bright center regions , and the x - ray photon indices near the @xmath0-ray peaks are roughly consistent with those in the tev energies in both objects ( @xmath94 for hess j1420 - 607 and @xmath95 for hess j1418 - 609 , @xcite ) . in addition to that , the difference of the photon index between the bright center and diffuse emission is @xmath96 . the smooth spectral steepening of the synchrotron x - ray emission likely reflects synchrotron burn - off of the accelerated electrons . such effects have been observed for some young pwne @xcite , but we now find an interesting example of the smoothly steepening of the synchrotron x - ray emission in a middle - aged pwn of k3/psr j1420@xmath976048 ( @xmath98 kyr ) . in order to estimate the mean magnetic field strength , we modeled the spectral energy distribution . since information about the spatial dependence of the tev emission is not available , we assumed a simple one - zone synchrotron + ic model as the first order approximation , in which a single population of relativistic electrons in tev energy range emit both x - rays through synchrotron radiation and tev @xmath0-rays through ic scattering off the cmb . the electrons energy distribution is formally assumed to be an exponentially cutoff power - law of the form @xmath99 , where @xmath100 . we do not specify the physical meaning of @xmath101 , which may account for both the maximum electron energy and the cooling break . this spectral form is invoked only for the purpose of an estimate of the magnetic field . * [ fig : sed ] shows the spectral energy distribution of hess j1420 - 607 and hess j1418 - 609 at the x - ray and @xmath0-ray bands ; the model curves are obtained with a combination of @xmath102 ( hess j1420 - 607 ) or @xmath103 ( hess j1418 - 609 ) and @xmath104 for both objects . the black thick lines overlaid on the _ suzaku _ data points indicate the absorbed power - law models with best - fit parameters as shown in table [ tab : fit ] . we note that the `` ring1 '' region mainly comes from bright pulsars while the `` reg1-reg2 '' region includes both emissions from pulsar and diffuse components . the deviation between the spectrum of the `` reg1-reg2 '' region and the model curve in higher x - ray energies comes from the pulsar s intrinsic hard spectrum since the intrinsic diffuse emission can be approximately estimated as `` reg1-reg2 '' - `` ring1 '' . * the magnetic fields can be estimated as @xmath105 g for hess j1420 - 607 and @xmath7 g for hess j1418 - 609 . these values are in agreement with the magnetic field strength that is expected for middle - aged pwne , which lends support to the pwn scenarios for both objects . the diffuse x - ray emission of k3/psr j1420 - 6048 has asymmetry ( a circular shape of the radius @xmath106 pc + a tail - like component ) and its tail - like component extends to the @xmath0-ray peak . on the other hand , the diffuse emission in the rabbit seems to be more concentrated near the bright central source with higher surface brightness and relatively symmetrical shape of @xmath107 pc . the angular sizes of the vhe @xmath0-ray sources are larger than those of the x - ray emission regions , e.g. @xmath108 pc ( hess j1420 - 607 ) and @xmath109 pc ( hess j1418 - 609 ) . for the estimated magnetic field of @xmath110 g , the energy loss rate due to synchrotron radiation is just comparable to the loss rate via ic scattering off the cmb in the thomson regime , which is given by @xmath111 the lifetime of the electrons emitting synchrotron x - rays at 1 kev is @xmath112 kyr for the estimated magnetic field strength . on the other hand , the energy loss timescale of the electrons emitting ic @xmath0-rays at 1 tev is @xmath113 kyr ( see e.g. , * ? ? ? the tev emitting electrons have a longer lifetime than the cooling time of the x - ray emitting electrons , which is shorter than the characteristic age of psr j1420@xmath976048 and probably also shorter than that of the pulsar in the rabbit . therefore the morphological differences between the two bands can be expected , though details may depend largely on the pwn evolution @xcite . according to @xcite , @xmath0-ray to x - ray energy flux ratio of pwne is proportional to the characteristic age . the energy flux ratios for both objects are @xmath114 for psr j1420 - 6048 and @xmath115 for the rabbit nebula ( @xmath116 ) values were taken from @xcite ) . if we assume that the characteristic ages for both objects are correct , i.e. @xmath25 kyr for psr j1420 - 6048 , and @xmath37 kyr for rabbit , the flux ratios roughly follow the mattana s relation . the high - energy electrons are thought to be transported via convection and/or diffusion due to the magnetic fields randomized at the termination shock . let us assume that diffusion is the dominant transport mechanism . indeed the pulsar wind slows down from relativistic to relatively low expansion velocities at some distance after the pulsar wind has been shocked . the diffusion coefficient can be estimated as @xmath117 where @xmath118 is the observed size of the synchrotron x - ray nebula . on the other hand , the diffusion coefficient can be written as @xmath119 where @xmath120 is the mean free path of scattering , parameterized by a gyrofactor @xmath121 and the electron gyroradius @xmath122 . by equating eq . ( 2 ) with eq . ( 3 ) , * we can estimate the gyrofactor as @xmath123 for both objects , which indicates that the quite efficient acceleration has occurred in these pwne . * interestingly , diffusion of electrons is slow , nearly at the the bohm limit of @xmath124 . the presumed configuration of the toroidal magnetic field in pwne means that the direction of the magnetic field in the post - shock flow should be essentially perpendicular to the radial flow direction . therefore , tev electrons can be confined quite efficiently in the absence of a radial field component . in this regard , the slow diffusion is not unexpected for pwne . if instead diffusion of electrons occurs along a radial magnetic field line , the magnetic field in the post - shock flow must be highly turbulent so that the bohm limit is realized . to further investigate diffusion processes , we need to incorporate spatial and temporal evolution of the physical parameters of the post - shock flow into morphological and spectral modeling . we expect that the next - generation cherenkov telescopes , cherenkov telescope array ( cta ) , will make it possible to perform spatially - resolved spectroscopy at the tev @xmath0-ray band , allowing us to constrain the spatial and temporal evolution of the pwne by combining x - ray and tev data ( see * ? ? ?	we report on the results from _ suzaku _ x - ray observations of the radio complex region called kookaburra , which includes two adjacent tev @xmath0-ray sources hess j1418 - 609 and hess j1420 - 607 . the _ suzaku _ observation revealed x - ray diffuse emission around a middle - aged pulsar psr j1420 - 6048 and a plausible pwn rabbit with elongated sizes of @xmath1 and @xmath2 , respectively . the peaks of the diffuse x - ray emission are located within the @xmath0-ray excess maps obtained by h.e.s.s . and the offsets from the @xmath0-ray peaks are @xmath3 for psr j1420 - 6048 and @xmath4 for rabbit . the x - ray spectra of the two sources were well reproduced by absorbed power - law models with @xmath5 . the spectral shapes tend to become softer according to the distance from the x - ray peaks . assuming the one zone electron emission model as the first order approximation , the ambient magnetic field strengths of hess j1420 - 607 and hess j1418 - 609 can be estimated as 3 @xmath6 g , and @xmath7 g , respectively . the x - ray spectral and spatial properties strongly support that both tev sources are pulsar wind nebulae , in which electrons and positrons accelerated at termination shocks of the pulsar winds are losing their energies via the synchrotron radiation and inverse compton scattering as they are transported outward .
last year s results reported by wmap3 @xcite,@xcite seem to confirm the inflationary paradigm with the spectral index @xmath0 and agree with the cosmic concordance model strengthening the case for dark energy . therefore it is an actual chalenge for the theory of fundamental interactions to accommodate any of the two ideas in realistic extensions of the standard model ( bsms ) . four dimensional supergravities stemming from string theories posses field dependent couplings controlled by vevs of moduli - fields which have flat or featureless classical scalar potential . these fields interact with gravity - strength couplings and appear to be natural candidates to play the role of the inflaton . in addition , they mix at the level of kinetic or potential energy with all matter - like fields present in the lagrangian , which makes them a part of the inflationary dynamics even if the designed would - be inflaton is not a modulus but rather a matter field . hence , the issue of generating supersymmetric inflation is intertwined with the issue of moduli stabilisation and it is a multi - field rather than a single field phenomenon , which may lead to important consequences . the presence of more than one active field makes the analysis of the dynamics more complicated . however , there are two crucial advantages of such an approach . firstly , it becomes easier to model simultaneously succesful inflation , cancellation of the post - inflationary vacuum energy and trapping of the moduli . secondly , the physics of inflationary fluctuations is markedly different in multi - field setups than in the standard single field inflationary models , which makes the realistic spectrum of fluctuations more natural and easier to generate . one interesting illustration of the possible role of additional scalars in generating realistic inflation is the case of the dilaton in heterotic string effective lagrangian @xcite . it is well known that a pair of condensates ( a racetrack ) can easily generate a scalar potential with a nontrivial weakly coupled minimum for both scalar components of the dilaton superfield . an additional difficulty consists in the necessity to cancel the post - inflationary cosmological constant . to simplify the inflationary dynamics , let us assume that in addition to the dilaton there exists another modulus , the volume modulus @xmath1 , with a no - scale khler potential , which automatically makes the tree - level scalar potential semi - positive definite . we assume in addition , that that no - scale modulus becomes stabilised in a different sector of the model , for instance with the help of non - trivial d - terms @xcite . let us note at this point that the two pehnomena the cancellation of the cosmological constant and generation of the inflationary period are not independent , as discussed for instance in @xcite , but we ingnore this complication in what follows . the weakly - coupled racetrack can not generate inflation . there are no regions where slow - roll parameters could be made sufficienly small simultaneously in all directions in ( two - dimensional ) field space . moreover , the exponential scalar potential is so steep along the real component of the dilaton , that the field which starts its evolution from arbitrary initial conditions set at very high initial temperatures `` overshoots '' the shallow weakly - coupled minimum and runs away towards the `` cold '' asymptotic minimum at vanishing gauge coupling . both problems can be solved if an additional superfield is present . in the case the mass , @xmath2 has a dependence on the vacuum expectation value , @xmath3 of a scalar field , @xmath4 it is clear that the condensation scale and hence the dilaton potential which determines the condensation scale will also depend on @xmath5 this can arise through the moduli dependence of the couplings involved in the mass generation after a stage of symmetry breakdown . for example a gauge - nonsinglet field , @xmath6 may get a mass from a coupling to a field , @xmath7 , when it acquires a vacuum expectation value through a yukawa coupling in the superpotential of the form @xmath8 . in general the coupling @xmath9 will depend on the complex structure moduli , @xmath10 and so the mass of the @xmath11 field will be moduli dependent , @xmath12 @xmath13 in what follows we will consider the implications of a very simple dependence of @xmath14 on the @xmath15 which is sufficient to illustrate how @xmath16 can provide an inflaton if it is a ( modulus ) field which has no potential other than that coming from the @xmath14 dependence of the condensation scale . in particular we take @xmath17 where @xmath18 is a mass coming from another sector of the theory , possibly also through a stage of symmetry breaking . we shall assume for simplicity a canonical kinetic term , @xmath19 for the modulus @xmath20 , although in practice the form of its khler potential may be more complicated . we are now in a position to write down the form of the superpotential corresponding to two gaugino condensates driven by two hidden sectors with gauge group @xmath21 and @xmath21 respectively . for simplicity we allow for a moduli dependence in the second condensate only . the race - track superpotential has the form @xmath22the coefficients @xmath23 are related to the remaining string thresholds @xmath24 by @xmath25 etc . whose moduli dependence we do not consider here , @xmath26 is related to the difference of beta functions before and after a stage of symmetry breaking at the scale @xmath14 . we will now argue that the racetrack potential has all the ingredients to meet the challenges posed by successful inflation . there are two main aspects to this . firstly the domain walls generated by the racetrack potential naturally satisfy the conditions needed for topological inflation . as a result there will be eternal inflation within the wall which sets the required initial conditions for a subsequent period of slow - roll inflation during which the observed density perturbations are generated . the second aspect is the existence of a saddle point(s ) close to which the potential is sufficiently flat to allow for slow - roll inflation in the weak coupling domain . as noted above this is not the case for the pure dilaton potential but does occur when one includes a simple moduli dependence . as pointed out by vilenkin @xcite and linde @xcite , `` topological inflation '' can occur within a domain wall separating two distinct vacua . the condition for this to happen is that the thickness of the wall should be larger than the local horizon at the location of the top of the domain wall ( we call this the ` coherence condition ' ) . in this case the initial conditions for slow - roll inflation are arranged by the dynamics of the domain wall which align the field configuration within the wall to minimise the overall energy . the formation of the domain wall is inevitable if one assumes chaotic initial conditions which populate both distinct vacuua and moreover walls extending over a horizon volume are topologically stable . although the core of the domain wall is stable due to the wall dynamics and is eternally inflating , the region around it is not . as a result there are continually produced regions of space in which the field value is initially close to that at the centre of the wall but which evolve to one or other of the two minima of the potential . if the shape of the potential near the wall is almost flat these regions will generate a further period of slow - roll inflation , at which time density perturbations will be produced . in the case of the racetrack potential the coherence condition necessary for topological inflation appears to be rather easily satisfied . this condition states that the physical width of the approximate domain wall interpolating between the minimum at infinity and the minimum corresponding to a finite coupling should be larger than the local horizon computed at the location of the top of the barrier that separates them . the width of the domain wall , @xmath27 , is such that the gradient energy stored in the wall equals its potential energy , @xmath28 , where @xmath29 . , @xmath30 being the real part of the dilaton . ] the fact that the racetrack potential readily generates topological inflation offers solutions to _ all _ the problems related to modular inflation . with chaotic initial conditions for the dilaton at the planck era the different vacuua will be populated because the height of the domain walls separating the minima is greater than the equilibrium temperature @xmath31 so thermal effects will not have time to drive the dilaton to large values . this avoids the initial thermal - roll problem . then there will be regions of space in which the dilaton rolls from the domain wall value into the minimum at finite @xmath30 , moving from larger to smaller values , and thus avoiding the thermal - roll problem . in fact once created the vacuum bag at finite @xmath30 is stable , because it can not move back into the core and over to the other vacuum the border of the inflating wall escapes exponentially fast so the respective region of space is trapped in the local vacuum . finally , if after inflation the reheat temperature is lower than the critical temperature @xmath32 the minimum may disappear due to thermal corrections in the hot gauge sector . ] the thermal effects will be too small to fill in the racetrack minimum at finite @xmath30 and the region of space in this minimum will remain there , thus avoiding a late thermal - roll problem . although the existence of topological inflation seems necessary for a viable inflationary model , by itself it is not sufficient to generate acceptable density perturbations . what is needed is a subsequent period of slow - roll inflation with the appropriate characteristics to generate an universe of the right size ( @xmath33 e - folds of inflation ) and density perturbations of the magnitude observed . in this subsection we present an example of a viable inflationary model in which the inflaton is the pseudo - goldstone boson associated with the phase @xmath34 of the field @xmath35 . [ f2.1 ] [ f2.2 ] [ f2.3 ] numerical analysis of the complete lagrangian in the @xmath36 hyperplane shows that typically there are inflationary solutions . a very nice example corresponds to the choice of parameters @xmath37 and @xmath38 . there is a weakly interacting minimum at @xmath39 and @xmath40 ( we remind the reader that @xmath41 and @xmath42 ) . the structure of the potential in the neighbourhood of @xmath43 is shown in figure 1 from which it may be seen that there is a maximum at @xmath44 there is a domain wall between the weakly interacting minimum and the non - interacting minimum at @xmath45 . as may be seen from figures 2 and 3 it has a saddle point at @xmath46 @xmath47 @xmath48 @xmath49 inflation occurs within the domain wall and there is further slow - roll inflation outside the wall as the wall inflates to a size not supported by the dynamics generating the wall . the initial conditions or this slow - roll deserves some comment . the @xmath30 field starts very close to the saddle point at @xmath50 . the same is true for the fields @xmath51 and @xmath52 which , at the saddle point , have masses larger than the hubble expansion parameter at this point . however the field @xmath34 is a pseudo - goldstone field and acquires a mass - squared proportional to @xmath53 as may be seen from figure 3 , since @xmath54 is small , the potential is very flat in the @xmath34 direction and the mass of @xmath34 is much smaller than the hubble expansion parameter . as a result the vev of @xmath34 during the eternal domain wall inflation undergoes a random walk about the saddle point and so its initial value can be far from saddle point . with these initial conditions it is now straightforward to determine the nature of the inflationary period after the fields emerge from the region of the domain wall . this corresponds to the roll of the fields , @xmath30 , @xmath55 and @xmath52 from the saddle point to the weakly interacting minimum , but allows for @xmath34 to be far from the saddle point . the @xmath56 @xmath51 and @xmath52 fields rapidly roll to their minima . however the gradient in the direction of the @xmath34 field is anomalously small due to the pseudo - goldstone nature of the field . quantitatively , in the neighbourhood of the weakly interacting minimum , we find a negative eigenvalue of the squared mass matrix corresponding to the phase @xmath34 , and its absolute value is about @xmath57 times smaller than the positive eigenvalues . this is much smaller than the hubble expansion parameter at the start of the roll and so the @xmath58 field indeed generates slow - roll inflation . the remaining degrees of freedom can be integrated out along the inflationary trajectory . inflation stops after about @xmath59 e - folds at @xmath60 and the pivot point corresponds to @xmath61 . the value of @xmath62 at this point is @xmath63 so the the spectral index is @xmath64 , consistent with the wmap3 value at @xmath65 . the agreement with the normalisation of the spectrum is also readily achieved ( we remind the reader that the expectation value of @xmath66 can be considered as a free parameter for the purpose of tuning the overall height of the inflationary potential , as it is fixed in a separate sector of the model ) . note that the ` running ' of the spectral index is very small , @xmath67 , hence probably undetectable . to summarise , the moduli dependent racetrack potential has a saddle point which lead to a phenomenologically acceptable period of slow - roll inflation with the inflaton being a component of the moduli . no fine - tuning of parameters is required and the initial conditions are set naturally by the first stage of topological inflation . after inflation there will be a period of reheat and the nature of this depends on the non - inflaton sector of the theory which we have not specified here . from the point of view of the racetrack potential the main constraint on this sector is that the reheat temperature should be less than @xmath32 to avoid the thermal roll problem . however @xmath32 is quite high , much higher than the maximum reheat temperature allowed by considerations of gravitino production , so we expect this constraint will be comfortably satisfied in any acceptable reheating model . the inflationary scheme described above encounters some problems when post - inflationary phenomenology is considered . first of all , the scale of the gravitino mass tends to be somewhat large with respect to the fermi scale when the scale of inflation is made close to the upper limit imposed by cobe normalisation . secondly , one might wish to rely on more traditional dynamics than topological inflation and topological trapping . in addition , in supersymmetric extensions of the standard model there are typically many quasi - flat directions involving combinations of scalars which belong to the matter rather than moduli sector . hence , it is natural to consider matter - scalar driven inflation . however , as already explained in the introduction , and discussed at length in @xcite , matter scalars always mix with moduli . it turns out that very often the curvature along the direction of the relevant modulus is so large , that it spoils successful inflation along the direction of the matter scalar - the fields acquire large velocity transverse to the would - be inflationary trajectory and inflation ends prematurely . in the paper @xcite various ways out of this rather generic problem have been discussed . it turns out that sometimes certain amount of tuning and creative model building are necessary to generate sufficiently long epoch of slow - roll inflation . the discussion of this issue is rather model dependent , and for more details and further references we recommend papers @xcite and @xcite . it is interesting to put aside fine details of the construction of multi - field inflationary models and ask for generic phenomena which occur in such a setup . the most interesting feature concerns quantum fluctuations during the inflationary period , which serve as primordial fluctuations triggereing formation of large scale structure . in the single - inflaton models fluctuations essentially do not evolve after horizon crossing , staying frozen until the second , post - inflationary horizon crossing , when the physical wavelength of a fluctuation with a given wave number becomes smaller than the frw horizon . however , when additional active inflatons are present , fluctuations in various fields get coupled and can evolve significantly after the first horizon crossing . in fact , this offers novel possibilities in shaping the spectrum of inflationary fluctuations . the point is that in the direction transverse to the classical inflationary trajectory the potential does nt have to be as flat as along the classical trajectory . hence , transverse fluctuations ( isocurvature ones ) do not need to have the spectrum as flat as that of the curvature fluctuations which borne as fluctuations in the ` momentary ' inflaton . if the coupling between both types of fluctuations is large enough , the isocurvature fluctuations can feed the curvature perturbations , and their not - so - flat spectrum becomes imprinted on the final spectrum of primordial fluctuations resulting from inflation . as the result , it becomes reasonably easy to obtain the scalar spectral index significantly smaller than 1 , as suggested by wmap3 . examples of such solutions are presented in @xcite . here we shall only illustrate the above statements . in the figure [ figcla ] examples of bent and curved classical inflationary _ , title="fig:",height=226 ] _ , title="fig:",height=226 ] _ , title="fig:",height=226 ] trajectories are given . all of them give rise to significant isocurvature perturbations . to be more specific let us consider the so - called roulette inflation model which has been investigated recently in @xcite ( see also @xcite ) . in the notation of @xcite this model can be effectively described by @xmath68 and @xmath69 where @xmath70 and @xmath71 , @xmath72 , @xmath73 are functions of the parameters of the underlying string model . a generic feature of the potential ( [ potbkpv ] ) is that it has an infinite number of minima arranged periodically in @xmath35 and a plateau for large values of @xmath55 , admitting a large variety of inflationary trajectories , which may end at different minima even if they originate from neighboring points in the field space hence the model has been dubbed _ roulette inflation_. in this work , we adopt the parameter set no . 1 ( in planck units : @xmath74 ; @xmath75 ; @xmath76 ; @xmath77 ; and @xmath78 ; @xmath79 ) from @xcite and choose the particular inflationary trajectory shown in figure [ figcla ] . for this trajectory , the factor @xmath80 is rather large , of the order @xmath81 , but the effect of the non - canonical kinetic terms is strongly suppressed by a very small value of @xmath82 on the plateau of the potential . the smallness of @xmath82 also suppresses the energy scale of inflation and one needs a smaller number of efolds than in the models described above . for definiteness , we assumed that there are @xmath83 efolds between the moment that the scale of interest crosses the hubble radius and the end of inflation . power spectrum of curvature fluctuations , @xmath84 power spectrum of isocurvature perturbations , normalized to the single - field result , respectively . the coupling @xmath85 between the curvature and isocurvature perturbations is also shown . ( full description can be found in @xcite . ) [ figby]_,height=340 ] the largest portion of the inflationary trajectory in this example lies on the plateau of the potential ( [ potbkpv ] ) , the slow - roll parameter @xmath82 is very small , which makes the direct impact of the non - canonicality negligible . the trajectory is , however , strongly curved in the field space and the interaction between the isocurvature and curvature modes is still important . as shown in @xcite one can accurately predict the spectra and correlations in the vicinity of the hubble crossing , with deviations on super - hubble scales resulting from the sourcing of the curvature perturbations by the isocurvature ones . eventually , most of the curvature perturbations arise through this effect . multi - field inflation , with more than a single field active during the inflationary period , is a rather generic phenomenon in unification theories based on supersymmetry and extra dimensions . this fact has several implications for inflationary model building and also for the post - inflationary dynamics . first of all , in the case of modular inflation one can use some of the active fields to trap run - away moduli via the topological inflation while the remaining ones could be used to create slow - roll inflation . a minimal example of such a scheme presented here following @xcite has several attractive features : * there is an initial period of topological inflation which sets the initial conditions for slow roll inflation and avoids the rapid roll and thermal roll problems usually associated with the racetrack potential . as a result there is no difficulty in having our universe settle in the weakly coupled minimum of the dilaton potential and not in the runaway non - interacting minimum . * the racetrack potential with simple moduli dependence has saddle points which lead to slow roll inflation capable of generating the observed density fluctuations with a spectral index smaller than but close to @xmath86 due to the initial period of topological inflation the initial conditions for the slow roll inflation are automatically set without fine tuning . another far - reaching feature of multi - field inflation is the presence of isocurvature perturbations . in some cases they may have a dramatic impact on the final spectrum of primordial fluctuations which trigger structure formation . quite often they easily lower the scalar spectral index below 1 , thus naturally reproducing the tendency implied by wmap3 data . 99 d. n. spergel _ et al . _ , `` wilkinson microwave anisotropy probe ( wmap ) three year results : implications for cosmology , '' arxiv : astro - ph/0603449 . w. h. kinney , e. w. kolb , a. melchiorri and a. riotto , `` inflation model constraints from the wilkinson microwave anisotropy probe three - year data , '' arxiv : astro - ph/0605338 . z. lalak , g. g. ross and s. sarkar , `` racetrack inflation and assisted moduli stabilisation , '' nucl . b * 766 * ( 2007 ) 1 [ arxiv : hep - th/0503178 ] . j. ellis , z. lalak , s. pokorski and k. turzynski , `` the price of wmap inflation in supergravity , '' jcap * 0610 * , 005 ( 2006 ) [ arxiv : hep - th/0606133 ] . a. vilenkin , `` topological inflation , '' phys.rev . 72 * , 3137 ( 1994 ) [ arxiv : hep - th/9402085 ] . a. d. linde and d. a. linde , `` topological defects as seeds for eternal inflation , '' phys . d * 50 * , 2456 ( 1994 ) [ arxiv : hep - th/9402115 ] .	moduli with flat or run - away classical potentials are generic in theories based on supersymmetry and extra dimensions . they mix between themselves and with matter fields in kinetic terms and in the nonperturbative superpotentials . as the result , interesting structure appears in the scalar potential which helps to stabilise and trap moduli and leads to multi - field inflation . the new and attractive feature of multi - inflationary setup are isocurvature perturbations which can modify in an interesting way the final spectrum of primordial fluctuations resulting from inflation .
lab on a chip technologies offer a range of unique opportunities for preparation and manipulation of molecular species . in particular , the compartmentalisation of biomolecules into monodisperse , micrometer - sized droplets allows for quantitative , high - throughput biochemical studies such as directed evolution,@xcite screening for reagents , reaction conditions or cells,@xcite as well as for the fabrication of designer emulsions and microgels.@xcite microdroplets can also allow the study of rare events , such as nucleation , and have thus enabled studies of the nucleation step of a@xmath0 aggregation@xcite as well as insulin amyloid growth.@xcite with the rapid development of microfluidic technologies , the need of ultra - sensitive detection methods becomes ever more pressing . a large fraction of present - day experiments rely on optical detection,@xcite with alternative strategies including , for instance , electrochemistry,@xcite mass spectrometry@xcite or raman spectroscopy.@xcite also , infrared ( ir ) spectroscopy techniques have been utilised to monitor the contents of microfluidic flows.@xcite here , we demonstrate an approach for performing off - line ir spectroscopy on the contents of single microdroplets with sub - micrometer spatial resolution . fourier transform infrared spectroscopy ( ftir ) is a key method for studying conformational properties of proteins and in particular for inferring their secondary structure.@xcite exposed to ir radiation , chemical bonds undergo vibrations such as stretching , bending and rotating . in the case of proteins , this leads to a spectrum characterised by a set of absorption features in the amide bands.@xcite thereby , the modes most commonly used to study the structural properties of polypeptides are the amide i , amide ii and amide iii bands . amide i arises mainly from stretching vibrations and is generally localised within 1690 - 1600 @xmath1 ; the exact band position is determined by the backbone conformation - in other words by the secondary structure of the protein . in contrast , amide ii originates from a combination of bending and stretching and is localised around 1580 - 1510 @xmath1 . it is still possible to associate the position of the band to the protein s secondary structure , but the fact that this band stems from a combination of two different modes makes this analysis less straightforward . finally , the amide iii band is a combination of many modes such as stretching , in - plane bending , stretching as well as bending and occurs in the range of 1300 - 1200 @xmath1 . in practice , @xmath2-helical structures have this band centred around 1654 @xmath1 , random coil proteins show a maximum around 1640 @xmath1 , and @xmath0-sheet - rich amyloidic aggregates exhibit an amide i maximum within 1610 - 1630 @xmath1.@xcite to achieve sub - micron spatial resolution for protein ir spectroscopy experiments , we utilised an thermomechanical detection technique based on atomic force microscopy ( afm ) : if an ir pulse at a given wavelength is absorbed by a sample , the local temperature rise leads to local thermal expansion.@xcite this deformation excites a mechanical resonance in the afm cantilever which is in contact with the analyte . afm detection of this temporary expansion of the scanned region therefore allows nanoscale resolution ir imaging and acquisition of local chemical spectra . simultaneously with the ir - absorption image , the system is able to scan topography ( with conventional contact mode ) and sample stiffness ( related to the frequency of the cantilever oscillations ) . in order to be able to reliably locate and distinguish the protein contents of individual microdroplets for performing nanoscale ir spectroscopy as well as to provide for an enormous amount of statistics , the droplets are aligned on a znse prism using a pre - patterned polymer stamp and dried overnight , as shown schematically in fig . [ fig : scheme](a)-(c ) . thereafter , spectra of single dried microdroplets containing monomeric and aggregated protein are acquired , and the ability to easily differentiate between the two demonstrates the efficacy of the presented approach . the capability of studying the contents of single microdroplets individually with a high - precision method paves the way for a wide range of experiments harnessing the advantages of microfluidics . for instance , combination with on - chip selection techniques@xcite that pre - screen for a predefined species within droplets could allow for specific in - depth investigation of analytes in the context of protein aggregation as well as directed evolution . in addition , it should be emphasised that the technique of aligning individual microfluidic droplets off - chip offers further possibilities for systematic ex - situ assays beyond ir spectroscopy or afm . in brief , micrometer - sized droplets of protein solutions in fluorinated oil are generated via a microfluidic droplet maker and deposited on a znse prism ( fig . [ fig : scheme](a ) ) . these droplets are then aligned on a grid using a patterned stamp of polydimethylsiloxane ( pdms ) as shown in fig . [ fig : scheme](b ) and dried overnight in a desiccator at room temperature or , alternatively , in an oven at @xmath3c . to measure the ir spectrum , the dried protein is heated locally using a laser and the resulting thermal expansion is determined using an afm tip , which is sketched in fig . [ fig : scheme](c ) . in the following , these steps are described in more detail . for the monomeric solution , lysozyme from chicken egg white ( sigma - aldrich , _ @xmath462970 _ ) is dissolved in deionised water at a concentration of 6 mg / ml . aggregates are formed by mixing 60 mg lysozyme with 1 @xmath5 sodium azide , 200 @xmath6l of 1 m hcl , 600 @xmath6l of 10 mm hcl , 200 @xmath6l of 2 m nacl and 5 @xmath6l of a preformed seed - fibril solution , filtering through 0.45 @xmath6 m pores , followed by incubation at @xmath3c for 24 h. this approach yields approximately micrometer - sized fibrils that form a gel - like structure when encapsuled as an aqueous droplet.@xcite as depicted in fig . [ fig : pics](a ) , droplets are generated using a microfluidic junction with a cross - section of @xmath7 , fabricated through a standard soft lithography approach.@xcite the protein solution is injected through the central arm at a flow rate of 50 @xmath6l / h , whereas fluorinated oil ( fluorinert fc40 , sigma - aldrich , _ @xmath4f9755 _ ) containing @xmath8 surfactant ( _ n , nbis(n - propyl)polyethylene oxide - bis(2-trifluoromethyl polyper fluoroethylene oxide ) amide _ ) is pumped through the side channels at 100 @xmath6l / h . these settings result in droplets with a diameter of approximately @xmath9 which are collected in a microcentrifuge tube . one ml of solution with droplets from either monomeric or aggregated protein , respectively , are pipetted onto opposite ends of the surface of an attenuated total internal reflection element prism made out of znse monocrystals . thereafter , a pre - patterned pdms stamp is pressed onto each drop of liquid in order to align the droplets to its imprinted grid . the grid consists of @xmath10 indents to capture individual droplets , with a spacing of @xmath11 , and is imprinted into pdms by soft lithography.@xcite this process is illustrated in figs . [ fig : scheme](a ) and ( b ) ; for simplicity only one instance of droplet solution shown . the prism - including the alignment polymer - is then stored for 15 h in a desiccator . alternatively , the devices can be dried at elevated temperatures - for instance , in an oven at @xmath3c at ambient pressures for 15 h. this , however , will lead to further aggregation inside the droplets upon drying . careful removal of the pdms slab yields a regular pattern of the dried contents of single microdroplets - in the present case protein - as shown in fig . [ fig : pics](b ) . afm images of droplets dried using these two approaches are presented in the supplementary information ( sec . [ sec : si ] ) . samples were scanned by a commercial nano - ir microscopy system ( anasys instruments ) with a line rate between 0.02 - 0.08 hz in contact mode . we used a silicon cantilever ( appnano ) with a nominal radius of 10 nm and a nominal spring constant of 0.5 n / m . all images have a resolution of 512x256 pixels . spectra were collected with a step width of @xmath12 within the range of 1200 - 1800@xmath13 at @xmath14 of the instrument s maximal laser power . all measurements were performed at room temperature . our results demonstrate that using the approach presented here , it is feasible to align tens to hundreds - if desired even thousands - of microdroplets , as shown in fig . [ fig : pics](b ) . such a regular deposition allows for reliable , systematic off - line assays including ultra - high precision analytical tools as , for instance , the nanoscale ir spectroscopy utilised in this work . figures [ fig : meas](a)-(d ) show how the aggregated lysozyme from an individual microdroplet can be analysed accurately . presented is the spatially resolved absorption of infrared radiation , determined via thermal expansion as measured by an afm tip . the wavenumbers are fixed at 1640 @xmath1 ( a ) , 1620 @xmath1 ( b ) , 1530 @xmath1 ( c ) , 1250 @xmath1 ( d ) , corresponding to two instances from the amide i band , one from the amide ii band , and one from the amide iii band , respectively . while all the four plots share their topographical features , it is readily apparent that absorption is higher in the amide i band than the amide ii and iii bands . moreover , the absorption in the amide iii band exhibits a stronger spatial dependence . this could be linked to the fact that this band is sensitive to different vibrational modes and therefore is influenced by local conformational changes accompanying the transition of protein from its monomeric form into aggregates.@xcite thus , it is possible to investigate the ir absorption behaviour of the contents of individual droplets locally . furthermore , the correlation with the height measurement from the afm scan in fig . [ fig : pics](c ) emphasises that the recorded absorption originates from the contents of a single microdroplet . the complete spectrum of the lysozyme aggregates in this droplet - averaged over 12 spectra recorded at different locations and smoothed by a savitzky - golay filter - is given by the dashed blue line in fig . [ fig : meas](e ) . as expected from the spatially resolved data , the nearly constant absorption observed in the amide i band - lightly peaked at approximately 1640 @xmath1 - is higher than in the amide ii and iii bands . remarkably , when comparing with a spectrum taken from a droplet containing only monomeric protein ( the solid red line is the smoothed average over 15 individual spectra ) , striking differences are apparent . first and foremost , the monomeric protein exhibits a sharp peak at around 1655 @xmath1 , due to the high @xmath2-helical content of lysozyme , and a shoulder at 1640 @xmath1 originating from random coils and @xmath0-sheets - all in good agreement with the structure of lysozyme@xcite and providing evidence that the secondary structural elements are largely unaffected by the gentle drying procedure . secondly , the amide ii band seems slightly shifted towards higher energies for the monomer , and thirdly , absorption in the amide iii band is significantly higher in monomeric than in aggregated protein . the shift of the amide i peak as well as the dramatic increase of absorption at 1620 @xmath1 are the typical signatures of the formation of amyloid - like cross-@xmath0 structure@xcite and have been studied extensively for the case of lysozyme.@xcite notably , our spectra from dried lysozyme correspond very well to measurements obtained in bulk solution.@xcite in fact , even if monomers already contain @xmath0-sheet domains their spectra differ from the amyloidic @xmath0-sheets and can be distinguished by the change in the location of the amide i peak.@xcite similarly , the position of the amide ii band is expected to shift towards lower wavenumbers if the secondary structure changes from predominantly @xmath2 helical to @xmath0-sheet.@xcite as the amide iii band has a significantly more complex origin , the differences in the monomeric and aggregated spectra are less directly explainable , but strong deviations are reasonable bearing in mind the extensive structural modification proteins undergo during aggregation . finally , comparison to a spectrum from a droplet containing initially monomeric protein that was dried at @xmath3c and ambient pressure for 15 h - such that aggregation can occur within the droplet - reveals that the monomeric features of the amide i and iii bands are lost upon fibrillation ( dash - dotted green line ; average over 21 spectra ) . note that due to the normalisation of the spectra to 1 the amide ii band seems more pronounced . nevertheless , the relative amplitudes of the amide ii and iii bands correspond very well to the spectra from the aggregates . for a znse prism that was covered with sio@xmath15 and heated with droplets containing monomeric lysozyme , we observed a similar increase of absorption in the amide i band but the peak in the amide iii band did not vanish , which may be an indication of partial aggregation ( see sec . [ sec : si ] ) . a technique for deposition and alignment of individual micrometer - sized droplets for their precise analysis using nanoscale spatially resolved ir spectroscopy was presented . off - stream alignment on a grid was achieved by means of a stamp with a patterned indentation on its surface . drying overnight fixes their content which is accessible upon removal of the polymer grid . subsequent high - precision measurement of local ir absorption demonstrates the power of this approach to probe structural transitions in ultra small volumes . spectra from droplets containing monomeric , aggregated and aggregating lysozyme were obtained and found to be readily distinguishable . in particular , the shift in the amide i band allowed us to identify an @xmath2-to-@xmath0 secondary structure transition which is associated with amyloid formation . while nanoscale ir spectroscopy represents a valuable analytic technique for the investigation of the contents of microfluidic droplets , the method of their alignment is not restricted to infrared spectroscopy . indeed , any technique requiring systematic ex - situ access to the microdroplets content is compatible with the demonstrated protocol . we thank pablo aran terol for fabricating the photolithography master for the microfluidic droplet maker and julius kirkegaard for help with the 3d sketches . financial support from the biotechnology and biological sciences research council ( bbsrc ) , the frances and augustus newman foundation , and the swiss national science foundation ( snf ) is gratefully acknowledged . several conditions for drying and analysing the droplets were investigated . namely , we have deposited microdroplets on pure znse prisms and sio@xmath15-coated znse prisms . drying was effected by either placing the prisms under vacuum at room temperature or in an oven at @xmath16 at ambient pressure for a time span of 15 h. the resulting morphologies varied significantly , and are summarised in fig . [ fig : afmscans ] . most importantly , droplets that have been heated overnight ( g - j ) exhibit a smaller size which may well be caused by the protein becoming more attached to the pdms stamp and therefore being removed when detaching the stamp . note that in the main work , only data from vacuum - dried droplets is presented . . ( b ) same as ( a ) , for a different droplet . the full scale height is @xmath17 . ( c ) droplet containing aggregates , deposited on a sio@xmath15-coated znse prism and dried for 15 h at room temperature under vacuum . the full scale height is @xmath17 . ( d ) same as ( c ) but for a different droplet . the full scale height is @xmath17 . ( e ) droplet containing monomeric lysozyme , deposited on a znse prism and dried for 15 h at room temperature under vacuum . the full scale height is @xmath18 . ( f ) droplet containing monomers , deposited on a sio@xmath15-coated znse prism and dried for 15 h at room temperature under vacuum . the full scale height is @xmath19 . ( g ) droplet containing aggregated lysozyme , deposited on a znse prism and dried for 15 h at @xmath16 at ambient pressure . the full scale height is @xmath20 . ( h ) droplet containing monomeric lysozyme , deposited on a znse prism and dried for 15 h at @xmath16 at ambient pressure . the full scale height is @xmath20 . ( i ) same as ( h ) , but for a different droplet . the full scale height is @xmath20 . ( j ) same as ( h ) , but for a different droplet . the full scale height is @xmath20.,scaledwidth=40.0% ] to demonstrate the repeatability of our approach , we present spectra of droplets containing monomeric and aggregated lysozyme dried under different conditions in figs . [ fig : spectra](a ) and ( b ) , respectively . specifically , we have investigated two different types of infrared - transparent prisms ( znse and sio@xmath15-coated znse ) and dried the droplets either at room temperature under vacuum or at @xmath16 at ambient pressure . for the monomeric protein , the vacuum - dried droplets on either substrate exhibit a distinct peak around 1655 @xmath1 as well as a sharp feature in the amide iii band which is pronounced more strongly for the plain znse prism . heating the droplets for 15 h to evaporate all solvents lead to a drastic increase of the absorption in the amide i band ( see black ellipse in fig . [ fig : spectra](a ) ) that corresponds very well to the spectra observed for aggregated protein . furthermore , in the case of the heated , sio@xmath15-coated prism the peak in the amide iii band also vanishes . considering the original approach for forming the aggregates , it is highly plausible that the conditions for drying the droplets at elevated temperatures are sufficient to cause at least partial aggregation of the initially monomeric protein - in particular since the evaporation of water continuously increases the protein concentration . the spectra of the droplets containing aggregated protein presented in fig . [ fig : spectra](b ) are very similar for all conditions and demonstrate the reliability of our approach . the most notable difference is that with the additional layer of sio@xmath15 on the surface of the prism the absorption at the high - energy side of the amide i band is increased relatively to the rest of the spectrum . the peak positions , however , are unchanged . also , the spectra from the three droplets on the same substrate ( solid , dashed and dotted blue lines ) indicate the quantitative variability between individual droplets with nominally the same contents . in both cases - that is monomeric and aggregated protein - some of the differences between the spectra from znse prisms and sio@xmath15-coated znse prisms may be explained by the different absorption behaviour on different substrates . indeed , in all instances of the coated prisms we observe stronger absorption towards higher wavenumbers . therefore , if protein aggregation from microdroplets is investigated , it is highly recommended to effect the drying under vacuum instead of at elevated temperature . furthermore , we find that our spectra exhibit sharper features on the pure znse prisms . given a droplet diameter of @xmath9 , the molar weight of lysozyme ( 14.3 kda ) and the concentration we have used for the monomers ( @xmath21 ) , each droplet contains as little as 50 pg or 3 fmol of protein . we can estimate the sensitivity of this approach via the minimal sample thickness ( recommended is @xmath22 , but we have obtained spectra from samples of @xmath23 height ) of our anasys nanoir platform ( http://www.anasysinstruments.com ) . to do so we evaluate the amount of analyte deposited on a square with length @xmath24 to a height @xmath25 using the inverse density of lysozyme ( @xmath26 ) . this yields @xmath27 and with @xmath24 being maximally of the order of @xmath28 the minimal concentration is around @xmath29 . while standard optical techniques may detect much smaller concentrations,@xcite nanoir provides a quite different set of information - namely nanometer spatially resolved ir spectra from picograms or sub - femtomoles of protein material . the time required to obtain a spectrum of a single droplet is typically of the order of a few minutes however , performing spatially resolved measurements may take much longer . therefore , nanoir can by no means be considered a high - throughput technique . however , nanoir may provide information on the contents of the droplets that is not accessible using traditional techniques and should rather be regarded as a method complimentary to on - line high - throughput techniques where measurements can be taken at khz rates at concentrations as low as tens of nm.@xcite	droplet microfluidics has emerged as a powerful platform allowing a large number of individual reactions to be carried out in spatially distinct microcompartments . due to their small size , however , the spectroscopic characterisation of species encapsulated in such systems remains challenging . in this paper , we demonstrate the acquisition of infrared spectra from single microdroplets containing aggregation - prone proteins . to this effect , droplets are generated in a microfluidic flow - focussing device and subsequently deposited in a square array onto a znse prism using a micro stamp . after drying , the solutes present in the droplets are illuminated locally by an infrared laser through the prism , and their thermal expansion upon absorption of infrared radiation is measured with an atomic force microscopy tip , granting nanoscale resolution . using this approach , we resolve structural differences in the amide bands of the spectra of monomeric and aggregated lysozyme from single microdroplets with picolitre volume .
many nonlinear phenomena appearing in physics are described by first - order quasilinear systems in both their hyperbolic and elliptic regions . these systems have been mostly studied in the case of two independent variables . however , until now , no satisfactory complete theory exists for those systems . for example , with the exception of certain particular results , the existence and uniqueness theorem for solutions of initial and boundary value problems is not available in general . such exceptions include the necessary and sufficient conditions for the temporal existence of smooth solutions of hyperbolic systems admitting conservation laws @xcite . solutions of hyperbolic quasilinear systems do not exist for an arbitrary period of time , even for smooth conditions . they usually blow up at the end of a finite time interval . in general , the first derivatives of the solution become unbounded after a finite time @xmath1 , and for a time @xmath2 , smooth solutions no longer exist . this well - known phenomenon is called the gradient catastrophe @xcite . it can even occur in situations where physical intuition would lead us to expect the existence of continuous solutions after time @xmath3 . in other words , the difficulty which appears here is to determine initial data for the quasilinear hyperbolic system which allows sufficiently large time interval before the gradient catastrophe occurs . it has been proved ( see _ e.g. _ @xcite ) that even for sufficiently small initial conditions , there exists a time interval @xmath4}$ ] in which the gradient catastrophe does not occur . in this interval , the problem of propagation and superposition of waves can be posed and solved by the method of characteristics . through this method , the existence , uniqueness and continuous dependence of the solution with respect to the initial conditions has been established by many authors ( see _ e.g. _ @xcite and references therein ) . the obtained results are significant in the sense that the solution is constructed in the domain where its existence is predicted . [ [ section ] ] the method of riemann invariants and its generalization , that is , the generalized method of characteristics ( gmc ) @xcite for hyperbolic quasilinear systems of equations in many dimensions , is a technique for obtaining certain classes of exact solutions representing riemann waves . these solutions are omnipresent for hyperbolic systems and constitute their elementary solutions . they are building blocks for the construction of more general solutions describing the superposition of many waves , @xmath0-waves ( solutions of rank @xmath0 ) , that are more interesting from the physical point of view @xcite . recently , the applicability of the conditional symmetry method ( csm ) has been extended to the construction of multiwave solutions obtained by the gmc @xcite . the csm is not limited to hyperbolic systems , but can also be applied to elliptic systems . the links between the two methods is an interesting problem which was addressed in @xcite . the adaptation of the gmc in the context of the analysis of the symmetry group of first - order quasilinear elliptic systems requires the introduction of complex integral elements instead of simple real integral elements ( which are used in the construction of solutions of hyperbolic systems @xcite ) . in particular , for first - order elliptic systems , we are interested in the construction of multimode solutions corresponding to nonlinear superpositions of elementary solutions ( simple modes ) . we demonstrate that the method of conditional symmetries @xcite is an efficient tool for reaching this goal . this approach is applied to hydrodynamic - type equations in their elliptic and hyperbolic regions and has been studied through both the gmc and the csm . this constitutes the subject of this work , which is a follow up of the paper @xcite . [ [ section-1 ] ] the proposed approach goes deeper into the algebraic aspect of first - order partial differential systems , which allows us to obtain some new results on their solvability . for this purpose we postulate a new form of solution of the initial system for which the matrix of derivatives of the unknown functions is expressed in terms of special orthogonal matrices , of the right - hand side of the partial differential system and of some characteristic vectors associated with the homogeneous part of this system . this decomposition of the matrix of derivatives is no longer in the specific form restricted to the sum of homogeneous and inhomogeneous integral elements as required by the gmc . in the latter case , when solutions exist , they represent a superposition of riemann waves which admit the freedom of @xmath0 arbitrary functions of one variable @xcite . for the postulate form of the solution the compatibility conditions are weakened in such a way that they allow us to obtain some larger classes of solutions including the one obtained by the gmc for the problem of superposition of waves or modes . [ [ section-2 ] ] the plan of this paper is as follows . section [ sec : gmc ] contains a brief description of the construction of riemann @xmath0 waves obtained by the gmc . section [ sec : ex : gmc ] contains examples of applications of the gmc to the equations of fluid dynamics in @xmath5 dimensions . in section [ sec:2 ] , we investigate and construct the multiwave solutions expressed in terms of riemann invariants which represent a generalization of the results obtained in section [ sec : gmc ] . section [ sec:3 ] contains a detailed account of the construction of multimode solutions for elliptic systems . sections [ sec:5 ] and [ sec:6 ] present an adaptation of the method presented in sections [ sec:2 ] and [ sec:3 ] for the case of underdetermined systems . several examples of inhomogeneous hydrodynamic - type equations are included in section [ sec:7 ] as illustrations of the theoretical results . section [ sec : finalremarks ] contains remarks and some suggestions regarding possible future developments . consider a first - order system of quasilinear autonomous partial differential equations ( pdes ) in @xmath6 independent variables [ eq : omr:1 ] ^i_(u ) u^_i = b^(u),=1, ,m,=1, ,q , i=1, ,p , where @xmath7 . we adopt the convention that repeated indices are summed unless one of them is in brackets . the system ( [ eq : omr:1 ] ) is an inhomogeneous one with coefficients depending on @xmath8 unknown functions @xmath9 . the euclidean space @xmath10 ( the space of independent variables @xmath11 ) is called the physical space , while the space @xmath12 ( the space of dependent variables ) is called the hodograph space . let us assume that the coefficients @xmath13 and @xmath14 are smooth real - valued functions of the real variables @xmath15 . we investigate the existence and the construction of solutions describing the propagation and nonlinear superposition of waves or modes that can be admitted by the system ( [ eq : omr:1 ] ) . such solutions are particularly interesting from the physical point of view because they cover a wide range of the wave phenomena arising in the presence of the external forces that can be observed in such domains as field theory , fluid dynamics , elasticity , _ etc_. these phenomena are described by systems of the form ( [ eq : omr:1 ] ) or systems that can be reduced to that form by introducing new unknown functions . the methodological approach assumed in this section is based on the generalized method of characteristics which was extensively developed in @xcite for homogeneous systems and generalized to inhomogeneous systems @xcite . the specific feature of this approach is an algebraic and geometric point of view . the algebraization of pdes was made possible by representing the general integral element as a linear combination of some special simple integral elements @xcite . [ [ section-3 ] ] the algebraization of the inhomogeneous system ( [ eq : omr:1 ] ) allows us to construct certain classes of solutions , which correspond to a superposition of @xmath0 simple waves together with a simple state solution ( as introduced in @xcite , see appendix ) . in this case , a specific constraint on the derivatives of a solution @xmath16 is postulated . we require that the jacobian matrix @xmath17 be the sum of homogeneous and inhomogeneous simple integral elements [ eq : gmc:2 ] = ^a_a^^a_i+^_0 ^0_i , where the summation is taken over the index @xmath18 which runs from 1 to @xmath0 and we have assumed that [ eq : gmc : ast ] _ ^i^_(a)_i^a=0,_^i_0^^0_i = b^ , + ^a_1^a_2^a_300<a_1<a_2<a_3k , and @xmath19here @xmath20 is treated as a function of @xmath21 . we suppose that we can find @xmath0 distinct characteristic vectors @xmath22 and @xmath0 linearly independent characteristic vectors @xmath23 which satisfy the homogenous equations ( [ eq : gmc : ast ] ) . we suppose also that the noncharacteristic vectors @xmath24 and @xmath25 obey the inhomogeneous equations ( [ eq : gmc : ast ] ) . the resulting solution of ( [ eq : gmc:2 ] ) is called a @xmath0-wave solution for the inhomogeneous hyperbolic system @xcite . under the above assumptions it was shown @xcite that the necessary and sufficient conditions for the existence of @xmath0-wave solutions of the system ( [ eq : omr:1 ] ) , subjected to conditions ( [ eq : gmc:2 ] ) , require the following constraints on vectors @xmath26 and @xmath27 , @xmath28 . the explicit parametrization of the integral surface @xmath29 in terms of riemann invariants @xmath30 is obtained by solving the system of pdes [ eq : gmc:3 ] = ^_n(u),n=0,1, ,k , with solution @xmath31 next , we look for the most general solution of the system of 1-forms @xmath32 for @xmath26 as functions of @xmath33 [ eq : gmc : dr ] dr^0=^0,dr^a=^a(x)^a(r),a=1, ,k . the compatibility conditions for the system ( [ eq : gmc : dr ] ) impose some restrictions on the wave vectors @xmath26 . namely , the system ( [ eq : gmc : dr ] ) has solutions ( is completely integrable ) if the following conditions are satisfied [ eq : gmc:5 ] \ { ^a,^n } , \ { ^n } , here we have denoted the vector space spanned by the vectors @xmath34 and @xmath26 by @xmath35 , while @xmath36 is proportional to @xmath26 . these relations are the necessary and sufficient conditions for the existence of solutions of the system ( [ eq : gmc : dr ] ) . they ensure that the set of solutions of the system ( [ eq : gmc : dr ] ) depends on @xmath0 arbitrary functions of one variable @xcite . finally , under the assumption that @xmath37 are linearly independent and satisfy ( [ eq : gmc:5 ] ) the @xmath0-wave solutions are obtained from the explicit parametrization of the integral surface @xmath38 . the quantities @xmath39 are implicitly defined as functions of @xmath40 by the solution of the system for certain functionally independent differentiable functions @xmath41 of @xmath33 _ i^a(r)x^i=^a(r),=_n^a(r)^(a)+^a_n(r)^(n),an , where @xmath42 and @xmath43 are given functions of @xmath44 . the physical interpretation of these solutions is that the profiles of simple waves related with the simple elements @xmath45 can be chosen in an arbitrary way , but the profile of a simple state related with the inhomogeneous element @xmath46 is somehow determined by ( [ eq : gmc : ast ] ) and ( [ eq : gmc:5 ] ) . these solutions represent some nonlinear superpositions of @xmath0 waves on a simple state @xcite . however , our present approach will go deeper into the geometrical aspects of compatibility conditions ( [ eq : gmc:3 ] ) and ( [ eq : gmc:5 ] ) by weakening them . so it will later enable us to obtain new results on the solvability of the problem of superposition of @xmath0-waves for inhomogeneous hyperbolic systems and extend this approach to elliptic systems . [ [ section-4 ] ] note that , in the case of the underdetermined system of equations ( [ eq : omr:1 ] ) ( _ i.e. _ when @xmath47 ) the real simple integral elements are determined by the equation _ ^i(u ) _ i ^=_s^i(u)_i_1^s+_t^i(u)_i _ 2^t = b^ , where @xmath48 , @xmath49 , @xmath50 and @xmath51 is the partition of the set @xmath52 into subsets containing @xmath53 and @xmath54 elements , respectively . the vector field @xmath55 has @xmath8 components , while @xmath56and the @xmath57 matrix @xmath58is a proper partition of the matrix @xmath59 . if the @xmath60 matrix @xmath61 is a nonsingular one , then we can determine the quantity @xmath62hence , there exists a bijective correspondence between @xmath63 and @xmath64 which determines a map on the set of simple integral elements and for which the domain is formed by these elements that have nonsingular matrices @xmath65this map is determined by the partition of the set @xmath66 . all possible partitions of this type determine an atlas composed of @xmath67 maps covering the set of real simple integral elements , where they are regular in the sense that @xmath68one can partition the set of simple integral elements in stratas numbered by the rank of matrices @xmath69 . regular elements form strata of the highest dimension . one can determine atlases on the other stratas analogously and solve the considered underdetermined system by the gmc . now we present some examples which illustrate the theoretical considerations presented in section [ sec : gmc ] . we discuss the classical equations of an ideal compressible nonviscous fluid placed in the presence of gravitational and coriolis forces . under these assumptions , the fluid dynamics system of equations in @xmath5 dimensions takes the form [ ex : gmc:1 ] & ( _ t+ ( ) ) + p=+ , + & _ t+()+=0 , + & ( p ^-k)_t+()(p^-k)=0 , where we have used the following notation : @xmath70 and @xmath6 are the density and the pressure of the fluid respectively , @xmath71 is the vector field of the fluid velocity , @xmath72 is the polytropic exponent and the external forces are the coriolis @xmath73 ( with angular velocity @xmath74 ) and the gravitational @xmath75 forces . the algebraic equations that determine simple integral elements for the equation ( [ ex : gmc:1 ] ) are of the form [ ex : gmc:2 ] & \|\| + ^p=+ , + & \|\| ^+=0 , + & \|\|(^p-^)=0 . here we have used the following notation : the space of unknown functions , the hodograph space @xmath76 , has coordinates @xmath77 . the corresponding elements of the tangent space @xmath78 are denoted by @xmath79 , where @xmath80 is associated with the velocity vector @xmath71 and @xmath81 are associated with the density @xmath70 and pressure @xmath6 , respectively . here we have denoted @xmath82 , where @xmath83 is the phase velocity and @xmath84 is the wave vector . we have replaced the derivatives of the unknown functions @xmath85 , @xmath86 , @xmath87 , @xmath88 , @xmath89 and @xmath90 in equation ( [ ex : gmc:1 ] ) by the simple elements @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath95 and @xmath96 , @xmath97 respectively . according to @xcite , we define the quantity [ ex : gmc : ast ] \|\|=_0 + , which physically describes the velocity of propagation of a wave relative to the fluid while @xmath83 describes the phase velocity of the considered wave @xmath98 , @xmath99 or @xmath100 . the equations ( [ ex : gmc:2 ] ) form a system of linear inhomogeneous algebraic equations , whose solutions @xmath101 and @xmath102 ( @xmath103 being the dual space of the classical space - time @xmath104 ) , determine the simple integral elements associated with the equation ( [ ex : gmc:1 ] ) . [ [ section-5 ] ] it follows from the analysis of the homogeneous system corresponding to ( [ ex : gmc:2 ] ) that there exist nontrivial solutions for the vector @xmath79 when the characteristic determinant of this system is equal to zero . thus we obtain the following condition ^3\|\|^3 ( ^2\|\|^2-\|\|^2 ) = 0 . so we obtain two types of homogeneous simple integral elements * the entropic simple element @xmath98 : + [ ex : gmc:4 ] = 0,=(- , ) , = ( ^,0 , ) , + with the condition @xmath105 . here @xmath106 , is an arbitrary function . * the acoustic simple element @xmath99 : + = & ( ^,^,- ) , + where @xmath107 is an arbitrary function . the nontrivial solutions for @xmath108 and @xmath109 of the inhomogeneous system ( [ ex : gmc:2 ] ) determine three types of the inhomogeneous simple integral elements of ( [ ex : gmc:1 ] ) , namely * the entropic inhomogeneous simple element @xmath110 + [ ex : gmc:6 ] = 0,_0^_0=(_,,(-)),^0=(-,- ) , + where @xmath111 is an arbitrary vector in @xmath112 and @xmath106 is an arbitrary function . * the acoustic inhomogeneous simple element @xmath113 + [ ex : gmc:7 ] \|\|=\|\| ( ) ^1/2,^0__0= ( \|\|- , ) , ( -)=0 , + _ 0^_0= ( _ , _ , ( \|\| ) ^-1 ( ( -)- _ ) ) , = 1 . + where @xmath106 is an arbitrary function . * the hydrodynamic inhomogeneous simple element @xmath114 + _ 0^=&(-,-k p , + & ) , + = & ( \|\|- , ) . here @xmath115 is different from 0 and @xmath116 , @xmath117 , and otherwise arbitrary . from the definition ( [ ex : gmc : ast ] ) , we obtain that the velocity associated with the solution of @xmath110 moves together with the fluid . for the case of acoustic solutions @xmath113 the velocity of the fluid is equal to the sound velocity : @xmath118 , while the hydrodynamic solution @xmath114 moves with any velocity other than the entropic velocity @xmath119 or acoustic velocity @xmath120 . [ [ section-6 ] ] we now consider the classes of solutions @xmath15 for which the matrix of the tangent mapping @xmath121 is the sum of one homogeneous and one inhomogeneous simple integral elements ( when @xmath122 in equation ( [ eq : gmc:2 ] ) ) [ ex : gmc:9 ] = _ 1^_i^1+_0^_i^0,^1 ^ 00 , where @xmath123 is treated as an arbitrary function of @xmath21 . according to @xcite the necessary and sufficient conditions for the existence of solutions of the system ( [ ex : gmc:9 ] ) require that the commutator of the vector fields @xmath124 and @xmath125 be a linear combination of these fields @xmath126}\in\operatorname{span}{\left\ { \gamma_0,\gamma_1 \right\}}.\]]this means that @xmath124 and @xmath125 constitute a holonomic system in the sense that there exists a parametrization of a surface @xmath127tangent to the vector field @xmath124 and @xmath125 such that [ ex : gmc:12 ] = _ 0 ( f(r^0,r^1 ) ) , = _ 1(f(r^0,r^1 ) ) holds . consequently the system ( [ ex : gmc:9 ] ) together with the assumption that @xmath124 and @xmath125 are linearly independent , require us to solve the following system of pdes [ ex : gmc:13 ] = _ i^0(r^0,r^1),=_i^1(r^0,r^1 ) . the involutivity condition for the system ( [ ex : gmc:13 ] ) has already been investigated in @xcite . these conditions lead to restrictions on the wave vectors @xmath128 and @xmath129 . it was shown that the system ( [ ex : gmc:13 ] ) is completely integrable if the following conditions are satisfied [ ex : gmc:14 ] = _ 0 ^0,=_1 ^1,=_0 ^0+_1 ^ 1,^0 ^ 10 , where the coefficients @xmath130 , @xmath131 , @xmath132 and @xmath133 are functions of @xmath134 and @xmath135 to be determined . consider the following two cases : [ [ section-7 ] ] * 1. + + + + when @xmath136 , the equations ( [ ex : gmc:14 ] ) can be integrated and the wave vectors @xmath129 and @xmath128 take the form @xmath137where the function @xmath138 is a solution of the equation @xmath139for a given function @xmath130 of @xmath134 and @xmath135 . all solutions of the system ( [ ex : gmc:13 ] ) can be obtained by solving the implicit relation with respect to the variables @xmath134 , @xmath135 and @xmath140 [ ex : gmc:15 ] ( r^1)x^i=&(r^1)+_0^r^0 ( -(s , r^1 ) ) ds , + x^i=&(r^1)-_0^r^0 ( -(s , r^1 ) ) ds , [ [ section-8 ] ] 2. + + + + when @xmath141 , the equations ( [ ex : gmc:14 ] ) can be integrated and the wave vectors @xmath129 and @xmath128 are @xmath142where @xmath143 and @xmath144 are arbitrary functions of their arguments , @xmath145 is an arbitrary function of @xmath146 and @xmath147 is a constant vector and where @xmath148 . the general integral of the system ( [ ex : gmc:13 ] ) can be obtained by solving the implicitly defined relations between the variables @xmath149 and @xmath140 [ ex : gmc:17b ] c_ix^i+c_0=_0^r^0(-(s))ds , + _ 0^r^0(s , r^1)(-(s))ds+a_i(r^1)x^i=(r^1 ) , where @xmath150 and @xmath151 are arbitrary functions of @xmath135 . if ( [ ex : gmc:17b ] ) can be solved , so that @xmath134 , @xmath135 and @xmath152 can be given as a graph over an open set @xmath153 , then the functions @xmath154 , determined from ( [ ex : gmc:12 ] ) , constitute an exact solution ( written in terms of the riemann invariants @xmath134 , @xmath135 ) of the inhomogeneous system ( [ ex : gmc:9 ] ) . a similar statement holds for the case @xmath136 and when one replaces the system ( [ ex : gmc:17b ] ) by the system ( [ ex : gmc:15 ] ) . [ [ section-9 ] ] superpositions of simple waves and a simple state , which are solutions of the system ( [ ex : gmc:1 ] ) and subjected to the differential constraints ( [ ex : gmc:9 ] ) are illustrated in table i , .the results of superpositions of simple waves determined by ( [ ex : gmc:1 ] ) and ( [ ex : gmc:9 ] ) [ cols="^,^,^,^",options="header " , ] are summarized as follows . where @xmath155 denotes that there exists a superposition written in terms of riemann invariants and @xmath156 denotes that there is no superposition . for convenience , we denote by @xmath157 , @xmath158 , @xmath159 , the solutions which result from nonlinear superpositions of waves and states associated with the given wave vectors @xmath160 and @xmath128 given by ( [ ex : gmc:4 ] ) and ( [ ex : gmc:6 ] ) or @xmath160 and @xmath128 given by ( [ ex : gmc:4 ] ) and ( [ ex : gmc:7 ] ) , respectively . we give several examples to illustrate the construction introduced in this section . [ [ the - entropic - simple - wave - and - the - entropic - state - mathfrakee_0 . ] ] 1. the entropic simple wave and the entropic state , @xmath157 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a solution exists , provided that @xmath161 , @xmath162 and @xmath163 , and it is given by = ( r^0),p = p(r^0 ) , + = -e^_0\|\|^-2c_0 + v_2(r)+ ( e^_0\|\|^-2(\|\|^2-c_0 ^ 2)^1/2 + 1 ) with riemann invariants c_0 t\|\|^-2(\|\|^2-c_0 ^ 2)^1/2-c_0\|\|^-2()+c_1=e^-_0r^0 , + e^-_0_0^r^0(r , r^1)dr+t + + a_1+a_3()=(r^1 ) , where @xmath164 , @xmath165 , @xmath166 and @xmath167 are arbitrary constants and @xmath168 , @xmath169 , @xmath170 , @xmath171 , @xmath172 and @xmath173 are arbitrary functions satisfying @xmath174 , @xmath175 and @xmath176 . [ [ section-10 ] ] the following additional entropic solution exists only if @xmath177 . it is given by = ( 0,0,g),=(0,0,1),=(r^0),p = p(r^0 ) , + ( r^1)=v_1(r^1)+ ( _ 0^r^1dr + -()_0^r^1dr+c)+v_3(r^1 ) , where the riemann invariants are given in implicit form by ( -v_1\|\|^2+\|\|^2_0^r^1dr+()^2_0^r^1dr + -c())t+ ( ( 1-v_3)+v_3()+v_1 ) = ( r^1)+r^0 , + ( ( ) ^2-\|\|^2 ) ( _ 1v_3-v_1_3)t+(-_3 + + _ 3()+_1)=(r^1 ) , where @xmath178 is an arbitrary constant , @xmath179 @xmath180 , @xmath181 , @xmath182 and @xmath183 are arbitrary functions such that @xmath184 and @xmath185 or @xmath186 . [ [ the - simple - entropic - wave - and - the - simple - acoustic - state - mathfrakea_0 . ] ] 2. the simple entropic wave and the simple acoustic state , @xmath158 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the case when @xmath187 and @xmath188 , the solution is given by = ( 0,0,g),=(_1,_2,0 ) , \|\|=1 , + = _ 0,p = p_0 , + = b_2(r^1 ) ( ^-1/2_0^r^0e^-(r)dr+b_1(r^1 ) ) + b_0 + + , where the riemann invariants are ( ^1/2-_1 b_0 ) t+_1+c_1=_0^r^0e^-(r)dr , + b_0t-=(r^1 ) , where @xmath189 @xmath180 , @xmath190 @xmath180 , @xmath191 and @xmath167 are arbitrary constants , @xmath192 , and @xmath193 , @xmath194 , @xmath195 and @xmath173 are arbitrary functions such that @xmath196 . [ [ the - simple - entropic - wave - and - the - simple - hydrodynamic - state - mathfrakeh_0 . ] ] 3. the simple entropic wave and the simple hydrodynamic state , @xmath197 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + when @xmath198 , @xmath199 , @xmath200 and @xmath201 , the solution is given by = ( 0,0,g),=(0,0,1 ) , + = ( r^1 ) , p = p_0 , + = -()+ ( _ 0^r^0e^-(r)dr+b(r^1 ) ) + ( ) , with riemann invariants = _ 0^r^0e^-(r)dr,=(c_1,c_2,0 ) , + + - a_2(r^1 ) ( ) ^2+a(r^1)t+(a_1(r^1 ) + + a_2(r^1)+a_3(r^1))=(r^1 ) , where @xmath202 , @xmath167 and @xmath203 are arbitrary constants ; @xmath204 , @xmath195 , @xmath173 , @xmath205 , @xmath206 and @xmath207 @xmath208 are arbitrary functions satisfying @xmath209 , @xmath210 , and @xmath211 or @xmath212 or @xmath213 . [ [ the - simple - acoustic - wave - and - the - simple - hydrodynamic - state - mathfrakah_0 . ] ] 4. the simple acoustic wave and the simple hydrodynamic state , @xmath214 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the case where @xmath215 and @xmath199 , @xmath216 , @xmath217 , @xmath188 , @xmath218 and @xmath219 , the solution is given by =( ( r^0,r^1)+ ( ) ) ^-1s(r^1),p = a^3 , + = ( r^0,r^1)+b_1 + ( _ 0^r^0e^-(r)dr+t_1 ) , where @xmath220 is given by the quadratic equation [ ex : gmc:27 ] ( + ( ) ) ^2+a(+())^-2s(r^1)^2+s(r^1)+s_1 + -(-t_1)_0^r^0e^-(r)dr+ ( _ 0^r^0e^-(r)dr ) ^2=0 , and the riemann invariants are given implicitly by ( ) t++c_1=_0^r^0e^-(r)dr , + t+((r^1))^-1_0^r^0e^-(r)(r , r^1)dr=(r^1 ) . here @xmath167 , @xmath221 , @xmath222 , @xmath223 and @xmath224 are arbitrary constants , @xmath225 is a constant vector such that @xmath226 ; @xmath117 ; @xmath195 , @xmath227 , @xmath173 and @xmath228 are any functions which satisfy ( [ ex : gmc:27 ] ) with @xmath229 , @xmath230 , @xmath231 , and @xmath232 . [ [ section-11 ] ] the generalized method of characteristics in the version presented in this paper proved to be a useful tool , since it led to many new interesting solutions . however , the superposition of simple waves and simple states is `` linear '' ( according to the classification presented in @xcite ) in the sense that the wave vector @xmath129 does not change the direction of propagation on a simple state . so , it seems to be worthwhile to try to extend this method and check its effectiveness for the case when the jacobian matrix @xmath17 is not necessarily expressible in terms of simple integral elements of the form ( [ eq : gmc:2 ] ) . this is , in short , the aim of the following sections . in this section we consider the possibility of generalizing the idea of riemann wave solutions of quasilinear systems to what may be considered a nonlinear superposition of riemann waves . the resulting solution will be called a riemann @xmath0-wave or simply a @xmath0-wave . this has already been discussed for the simple wave in @xcite . we suppose that the system is well - determined ( @xmath233 ) and that it constitutes a system of @xmath8 equations and of @xmath8 dependent variables @xmath234 in terms of @xmath6 independent variables @xmath235 . we look for a solution of system ( [ eq : omr:1 ] ) of the form [ eq : omr:2 ] u = f(r^1(x , u ) , , r^k(x , u)),r^a(x , u)=^a_i(u)x^i , a=1 , , k , where the functions @xmath236 are the riemann invariants associated respectively with the real linearly independent wave vectors @xmath237 which satisfy the wave relation specified below . the jacobian matrix @xmath238 takes the form [ eq : omr:3 ] u=(u^_i)=^-1^qp,=i_q-^qq , where we use the notation [ eq : omr:4 ] =( ) ^qk,= ( ^a_i ) ^kp , + =( ) = x^i^kq , @xmath239 ; @xmath240 ; @xmath241 ; @xmath242 . substituting ( [ eq : omr:3 ] ) into the system ( [ eq : omr:1 ] ) we obtain [ eq : omr:6 ] ^i _ ( ^-1 ) ^_^a_i = b^ or , in matrix form , [ eq : omr:7 ] ( ^i^a_i ) ^-1=b , where @xmath243 and @xmath244 . [ [ section-12 ] ] let us introduce @xmath0 real scalar functions @xmath245 , @xmath0 matrices @xmath246 and @xmath0 characteristic vectors @xmath247 such that [ eq : omr:8 ] ^-1=_(a)l_ab+_a , a=1, ,k , [ eq : omr:10 ] ^i_i^a_a=0 . note that the vectors @xmath248 are characteristic vectors of the homogeneous part of equation ( [ eq : omr:7 ] ) . next , we eliminate the quantities @xmath249 from equation ( [ eq : omr:7 ] ) through the relations ( [ eq : omr:8 ] ) . we find an algebraic relation for the quantities @xmath250 , @xmath34 and @xmath251 [ eq : omr:9 ] _ a=1^k ( ( _ a ( ^i_i^a ) l_a ) -i_q ) b=0 , the relation ( [ eq : omr:9 ] ) represents a constraint on the form of the scalar functions @xmath250 , of the wave vectors @xmath34 and of the rotation matrices @xmath251 . we multiply equation ( [ eq : omr:8 ] ) on the left by the matrix @xmath252 and then introduce the explicit expression for @xmath252 given in ( [ eq : omr:3 ] ) . next , we solve the resulting equations for the matrices @xmath253 in order to obtain [ eq : omr:11 ] =( _ b^+ ) ( i_k+(_b^+ ) ) ^-1 , where we have introduced the following notation @xmath254we assume that the @xmath255 dimensional matrix @xmath256 is invertible . in order to determine the conditions for which the system ( [ eq : omr:11 ] ) is well defined in the sense that it represents a system for the functions @xmath257 in terms of the invariants @xmath258 only , it is convenient to define a set of vector fields on the space @xmath259 [ eq : omr:14 ] x_a=^i_a(u)_x^i , a=1, ,p - k , with the property [ eq : omr:15 ] ^i_a_i ^a=0,a=1, ,p - k , a=1, ,k . applying the vector fields ( [ eq : omr:14 ] ) to the system ( [ eq : omr:11 ] ) and taking into account that @xmath260 annihilates all functions of the invariants @xmath258 , we find the following constraints [ eq : omr:16 ] x_a ( ( _ b^+ ) ( i_k+(_b^+ ) ) ^-1 ) = 0 , a=1 , , p - k . in conclusion , if the conditions ( [ eq : omr:10 ] ) , ( [ eq : omr:9 ] ) and ( [ eq : omr:16 ] ) are satisfied by an appropriate choice of the functions @xmath250 , @xmath34 , @xmath248 and @xmath251 , then a solution of system ( [ eq : omr:11 ] ) represents a multiwave solution of the hyperbolic system ( [ eq : omr:1 ] ) of the proposed form ( [ eq : omr:2 ] ) . it should be noted that these conditions are sufficient but not necessary . [ [ section-13 ] ] let us compare the proposed approach with the gmc . if we suppose that there exists a solution of the form ( [ eq : omr:1 ] ) such that the matrix @xmath252 is invertible , and that the vector @xmath248 obeys the wave relation ( [ eq : omr:10 ] ) , then we can determine a function @xmath250 and a rotation matrix @xmath251 in such a way that the relations ( [ eq : omr:8 ] ) are satisfied . indeed , since the vectors @xmath261 and @xmath262 have the same dimension , there exists a bijective transformation between these two vectors . this transformation is determined by the rotation matrix @xmath251 and the scalar function @xmath250 . consequently , the jacobian matrix of a solution takes the form [ eq : r : ast ] & = _ a ( _ a l^_a+_a^)_i^a + & = _ a ( _ a^(x , u)+_a^(x , u ) ) _ i^a(u ) , where @xmath263 so the jacobian matrix of a solution admits the following decomposition [ eq : r:2 ] = _ 1^^1_i++^_k^k_i , or equivalently , ( [ eq : r:2 ] ) written in its matrix form , is @xmath264where we have used the notation @xmath265 note that , in general , conditions ( [ eq : r : ast ] ) are weaker than the differential constraints ( [ eq : gmc:2 ] ) required by the gmc , since the latter are submitted to the wave relation ( [ eq : gmc : ast ] ) and differential constraints ( [ eq : gmc:3 ] ) and ( [ eq : gmc:5 ] ) . indeed , ( [ eq : omr:8 ] ) implies , that all first - order derivatives of @xmath15 with respect to @xmath140 are decomposable in the form ( [ eq : r:2 ] ) on some open domain @xmath266 , where @xmath267 are real - valued matrix functions defined on the first jet space @xmath268 . this fact allows us to ease restrictions imposed on the initial data at @xmath269 for hyperbolic systems . thus we are able to consider more diverse configurations of waves involved in superpositions ( described by inhomogeneous systems ) than in the gmc case . [ [ section-14 ] ] we end this section with two remarks : in the case where the matrices @xmath270 and @xmath271 depend only on the dependent variables @xmath15 , conditions ( [ eq : omr:16 ] ) simplify to @xmath272 which is equivalent to @xmath273}=0.\]]considering the notation ( [ eq : omr:4 ] ) , we find the sufficient condition @xmath274for which the system ( [ eq : omr:11 ] ) is well defined in the sense that the right - hand side is expressed as a function of the invariants @xmath275 only . more generally , a sufficient condition for the system ( [ eq : omr:11 ] ) to be expressible in terms of @xmath33 is that the matrix @xmath276 be proportional to the matrix @xmath277 . this means that there exists a scalar function @xmath278 such that @xmath279where @xmath277 is defined by ( [ eq : omr:4 ] ) . * if @xmath280 , there are no orthogonality conditions ( [ eq : omr:15 ] ) and therefore the conditions ( [ eq : omr:16 ] ) are no longer necessary in order to ensure that the system ( [ eq : omr:11 ] ) is well defined . in this case however , the linear independence of the wave vectors @xmath34 ensures that the relations @xmath281 can be inverted to obtain the @xmath140 in terms of the @xmath258 . consequently , the independent variables @xmath140 can be eliminated from the system ( [ eq : omr:11 ] ) , which is then expressed in terms of @xmath282 and @xmath33 only . [ sec:3 ] the algebraization that has been performed in @xcite for the elliptic system ( [ eq : omr:1 ] ) allows us to construct more general classes of solutions , namely the @xmath0-multimode solutions ( a superposition of @xmath0 simple mode solutions ) . we look for real solutions of the form [ eq : mm:1 ] u = f(r^1(x , u), ,r^k(x , u ) , \|r^1(x , u), ,\|r^k(x , u ) ) , + r^a(x , u)=_i^a(u)x^i,\|r^a(x , u)=\|_i^a(u)x^i , a=1, ,k . the complex - valued wave vectors @xmath34 and their complex conjugates @xmath283 are linearly independent . the jacobian matrix @xmath238 of a real - valued solution ( [ eq : mm:1 ] ) in terms of riemann invariants takes the form [ eq : mm:2 ] u=^-1 ( ^a+\|^a ) = ^-1 ( + \| ) , where [ eq : mm:4 ] = i_q--^qq,= ( ) ^qk,= ( _ i^a ) ^kp , where @xmath284 and @xmath285 are complex conjugates of @xmath286 and @xmath277 respectively . we introduce the jacobian matrix ( [ eq : mm:2 ] ) into the system ( [ eq : omr:1 ] ) , which results in the following system written in matrix form [ eq : mm:5 ] ^i ( _ i^a^-1+\|_i^a^-1 ) = b . we now define the complex - valued scalar and vector functions , @xmath245 and @xmath287 respectively , as well as the matrices @xmath288 such that [ eq : mm:6 ] ^-1=_(a)l_ab+_a,^-1=\|_(a)\|l_a b+\|_a , a=1, ,k , where @xmath289 , @xmath290 and @xmath291 are the complex conjugates of @xmath250 , @xmath248 and @xmath251 respectively . substituting the expressions ( [ eq : mm:6 ] ) into the system ( [ eq : mm:5 ] ) and assuming that the characteristic vectors @xmath248 satisfy [ eq : mm:7 ] ^i ( _ i^a_a+\|_i^a\|_a ) = 0 , we obtain the algebraic condition [ eq : mm:8 ] ( ( _ a^i ( _ a_i^al_a+\|_a\|_i^a\|l_a ) ) -i_q ) b=0 , on functions @xmath250 , @xmath34 and @xmath251 , @xmath241 and their complex conjugates . multiplying each of the equations ( [ eq : mm:6 ] ) on the left by @xmath252 and replacing the explicit expression ( [ eq : mm:4 ] ) for the matrix @xmath252 , we obtain the following pdes for the function @xmath282 [ eq : mm:9 ] & =( i_q-- ) ( _ ( a)l_ab+_a ) , + & =( i_q-- ) ( \|_(a)\|l_ab+\|_a ) . + solving the system ( [ eq : mm:9 ] ) algebraically for the matrices @xmath286 and @xmath284 , we find the system in its canonical form [ eq : mm:10 ] & =( i_q- ) ( _ b^+ ) ( i_k+ ) ^-1 , + & =( i_q- ) ( _ b^+\| ) ( i_k+ ) ^-1 , + where we have used the notations [ eq : mm : s ] _ =( _ ( a)l^_a ) ^qk,\|_= ( \|_(a)\|l^_a ) ^qk + s= ( _ b^+ ) ( i_k+ ) ^-1^qk,= ( ^_a ) ^qk and @xmath292 is its complex conjugate . using ( [ eq : mm : s ] ) , we can write system ( [ eq : mm:10 ] ) in the form [ eq : mm:11 ] & =( s- \|ss ) ( i_k+\|ss ) ^-1 , + & =( \|s- s\|s ) ( i_k+s\|s ) ^-1 . in order to verify that the system ( [ eq : mm:11 ] ) is well defined in the sense that the right - hand sides are expressible as functions of @xmath33 and @xmath293 only , we apply the vector fields [ eq : mm:12 ] x_a=^i_a(u)_x^i , i=1, ,p , a=1, ,p-2k , to the relations ( [ eq : mm:11 ] ) , where the coefficients @xmath294 satisfy the orthogonality conditions [ eq : mm:13 ] ^i_a(u)^a_i=0,^i_a(u)\|^a_i=0,a=1, ,k . as a result , we obtain the conditions [ eq : mm:14 ] x_a&=0,a=1 , , p-2k , + x_a&=0 , which are necessary and sufficient for the system ( [ eq : mm:11 ] ) to be well - defined in terms of the riemann invariants @xmath33 and @xmath293 . in conclusion , if the conditions ( [ eq : mm:7 ] ) , ( [ eq : mm:8 ] ) and ( [ eq : mm:14 ] ) are satisfied , then integrating ( [ eq : mm:11 ] ) , we obtain a solution describing a multimode solution of the system ( [ eq : omr:1 ] ) . in sections [ sec:2 ] and [ sec : modemul ] , we established a method allowing the construction of solutions expressed in terms of riemann invariants by restricting ourselves to well - determined systems ( @xmath295 ) . the proposed method can be modified so that it can be applied to underdetermined systems . consider the first - order quasilinear hyperbolic system of pdes ( [ eq : omr:1 ] ) , where the number of equations @xmath296 , and @xmath8 is the number of dependent variables . the expression for the jacobian matrix given in equation ( [ eq : omr:3 ] ) remains valid with the matrices @xmath286 , @xmath297 and @xmath277 defined in equation ( [ eq : omr:4 ] ) . therefore , by substituting the jacobian matrix into the system ( [ eq : omr:1 ] ) , we obtain equation ( [ eq : omr:6 ] ) for the case of underdetermined systems . in the well - determined case , the vectors @xmath298 and @xmath262 were of equal dimensions , which allowed us to express the vectors @xmath298 in terms of @xmath262 by the introduction of the scalar functions @xmath245 and the matrices @xmath246 . since the system is now underdetermined , the vectors @xmath298 can be expressed in terms of @xmath262 through certain rectangular matrices @xmath299 which are of dimension @xmath300 . the matrices @xmath299 are constructed from @xmath60 matrices @xmath301 , where @xmath53 is the number of equations in ( [ eq : omr:1 ] ) , and are defined by [ eq : omrsd:1 ] p_a=^_(a)(x , u)m_l^_a(x , u),a=1, ,k , where the @xmath302 scalar functions @xmath303 are real - valued and the constant matrices @xmath304 are defined by @xmath305the vectors @xmath306 can then be written as [ eq : omrsd:3 ] ^-1=p_a+_a , a=1, ,k , where the characteristic vectors @xmath307 depend on the independent variables @xmath21 and the dependent variables @xmath15 and obey the relation [ eq : omrsd:4 ] ^i_i^a_a , a=1, ,k , where @xmath308 . substituting the relations ( [ eq : omrsd:3 ] ) into equation ( [ eq : omr:7 ] ) , we obtain the algebraic relation [ eq : omrsd:5 ] ( ^i_i^ap_a - i_m ) b=0 , where @xmath309 is the identity matrix of dimension @xmath60 . in order to obtain the reduced system of pdes for the functions @xmath282 in terms of @xmath258 , we proceed in a way similar to that for the multiwave solution presented in section iii . if we suppose that [ eq : omrsd:6 ] ( i_k+ ) 0 , where @xmath310 , @xmath311 , the reduced system takes the form [ eq : omrsd:7 ] =( _ b^+ ) ( i_k+ ) ^-1 . as in the well - determined case of section [ sec:2 ] , the reduced system ( [ eq : omrsd:7 ] ) is well defined if the vector fields ( [ eq : omr:14 ] ) , satisfying the orthogonality conditions ( [ eq : omr:15 ] ) , cancel out the right - hand side of equation ( [ eq : omrsd:7]),_i.e . _ [ eq : omrsd:8 ] x_a=0 . in summary , if @xmath0 wave vectors @xmath34 and @xmath0 matrices @xmath299 , defined by ( [ eq : omrsd:1 ] ) , satisfy the relation ( [ eq : omrsd:5 ] ) , @xmath0 characteristic vectors @xmath248 satisfy the equation ( [ eq : omrsd:4 ] ) , and if the condition ( [ eq : omrsd:8 ] ) is fulfilled , then the reduced system ( [ eq : omrsd:7 ] ) is well defined for @xmath282 in terms of the riemann invariants @xmath258 . moreover , any solution of the system ( [ eq : omrsd:7 ] ) provides us with a multiwave solution of ( [ eq : omr:1 ] ) of the proposed form ( [ eq : omr:2 ] ) . in analogy with the case of multiwaves presented above , we can consider underdetermined elliptic systems ( [ eq : omr:1 ] ) in order to obtain real multimode - type solutions in the form [ eq : mmsd:1 ] u = f(r^1(x , u), ,r^k(x , u),\|r^1(x , u), ,\|r^k(x , u ) ) , + r^a(x , u)=^a_i x^i,\|r^a(x , u)=\|^a_i x^i , a=1, k , where the @xmath0 linearly independent wave vectors @xmath34 are complex - valued and @xmath283 are their complex conjugates . the jacobian matrix takes the form ( [ eq : mm:2 ] ) with matrices @xmath252 , @xmath312 and @xmath277 defined in equation ( [ eq : mm:4 ] ) , and so relation ( [ eq : mm:5 ] ) is also satisfied in the underdetermined case . the difference between the present case and that of multiwaves presented above in section [ sec:5 ] is that the functions @xmath313 and the matrices @xmath314 are complex - valued functions . so , the relations ( [ eq : omrsd:3 ] ) have to be complemented by their complex conjugate relations [ eq : mmsd:2 ] ^-1=\|p_a+\|_a , a=1, ,k , where @xmath315 are the complex conjugate matrices of the @xmath299 defined in ( [ eq : omrsd:1 ] ) and the characteristic vectors @xmath290 are complex conjugates of the vectors @xmath248 which , together , satisfy the relation [ eq : mmsd:3 ] ^i ( ^a_i_a+\|^a_i\|_a ) = 0 making use of ( [ eq : mmsd:2 ] ) and eliminating @xmath316 in equation ( [ eq : mm:5 ] ) we obtain the algebraic condition [ eq : mmsd:4 ] ( ^i ( ^a_i p_a+\|^a_i\|p_a ) + i_m ) b=0 . hence , the system of pdes for @xmath282 in terms of the @xmath258 and @xmath317 takes the form [ eq : mmsd:5 ] = & ( i_q-\| ) ( i_k+\| ) ^-1 , + = & ( i_q- ) \| ( i_k+\| ) ^-1 , + where @xmath318by analogy with the previous cases in section [ sec:5 ] , the system ( [ eq : mmsd:5 ] ) is well defined if the conditions [ eq : mmsd:6 ] x_a=&0 , + x_a=&0 , + are satisfied , with the vector fields @xmath260 defined by ( [ eq : mm:12 ] ) and ( [ eq : mm:13 ] ) . [ [ section-15 ] ] thus we have demonstrated that , if @xmath319 wave vectors @xmath34 and @xmath283 , and @xmath319 matrices @xmath299 and @xmath315 , defined by ( [ eq : omrsd:1 ] ) , satisfy the relation ( [ eq : mmsd:4 ] ) , @xmath319 characteristic vectors @xmath248 and @xmath290 satisfy the equation ( [ eq : mmsd:3 ] ) and if the conditions ( [ eq : mmsd:6 ] ) are fulfilled , then the reduced system ( [ eq : mmsd:5 ] ) is well defined for @xmath282 in terms of the riemann invariants @xmath258 . moreover , any solution of the system ( [ eq : mmsd:5 ] ) provides a multimode solution of the system ( [ eq : omr:1 ] ) of the proposed form ( [ eq : mmsd:1 ] ) . now we illustrate the theoretical considerations presented in sections [ sec:2]-[sec:6 ] with various cases of inhomogeneous hydrodynamic - type systems . [ [ section-16 ] ] * 1 . * + + + + consider a system of three equations for three unknown functions @xmath320 of the four independent variables @xmath321 , [ ex1:eq:1 ] u_t&+a_1 ( u_x+v_y+w_z ) + a_3 ( u_z - w_x)-a_2 ( v_x - u_y)=b_1(u , v , w ) , + v_t&+a_2 ( u_x+v_y+w_z)+ a_3 ( v_x - w_y)+a_1(v_x - u_y)=b_2(u , v , w ) , + w_t&+a_3 ( u_x+v_y+w_z ) - a_2 ( v_z - w_y)- a_1(u_z - w_x)=b_3(u , v , w ) , where @xmath322 are the components of the velocity along the @xmath21 , @xmath323 , @xmath324 , axes respectively and the @xmath325 , @xmath326 , are real constants . in matrix form , the system ( [ ex1:eq:1 ] ) can be written as @xmath327where u=(u , v , w)^t:^4 ^ 3,b=(b_1,b_2,b_3)^t , + ^1= ( ccc a_1 & -a_2 & -a_3 + a_2 & a_1 & 0 + a_3 & 0 & a_1 ) , ^2= ( ccc a_2 & a_1 & 0 + -a_1 & a_2 & -a_3 + 0 & a_3 & a_2 ) , ^3= ( ccc a_3 & 0 & a_1 + 0 & a_3 & a_2 + -a_1 & -a_2 & a_3 ) . we consider the possibility of constructing mixed wave type solutions which may be nonlinear superpositions of a simple mode with a simple riemann wave and we determine the admissible forms of the right - hand side @xmath262 of the system ( [ ex1:eq:1 ] ) . this solution can be expressed in terms of a real - valued invariant @xmath328 , of a complex - valued invariant @xmath329 and its complex conjugate @xmath330 . the invariants @xmath328 , @xmath329 and @xmath330 are associated with the real wave vector @xmath331 , the complex - valued wave vector @xmath332 and its complex conjugate @xmath333 , respectively . so , we look for solutions of the form @xmath334according to section [ sec:2 ] , the wave vectors @xmath335 , @xmath160 and @xmath333 and the special orthogonal matrices @xmath251 have to satisfy the algebraic condition ( [ eq : mm:8 ] ) , which , in the mixed case , takes the form [ ex1:eq:4 ] ( & _ 1 ( _ 0 i_3+_1 ^1+_2 ^2+_3 ^3 ) l_1 + _ 2(_0 i_3+_1 ^1+_2 ^2+_3 ^3)l_2 + & + \|_2(\|_0 i_3+\|_1 ^1+\|_2 ^2+\|_3 ^3)\|l_2-i_3)b=0 , where @xmath336 and the special orthogonal matrix @xmath337 are real - valued functions , but @xmath338 and the special orthogonal matrix @xmath339 are complex - valued functions . the wave relation ( which is here a mixed case of ( [ eq : omr:10 ] ) and ( [ eq : mm:7 ] ) ) for the real characteristic vector @xmath340 and the complex characteristic vector @xmath341 and its complex conjugate @xmath342 can then be written as [ ex1:eq:5 ] & ( _ 0 i_3+_1 ^1+_2 ^2+_3 ^3 ) _ 1 + ( _ 0 i_3+_1 ^1+_2 ^2+_3 ^3 ) _ 2 + & + ( \|_0 i_3+\|_1 ^1+\|_2 ^2+\|_3 ^3 ) \|_2=0 . the algebraic conditions ( [ ex1:eq:4 ] ) and ( [ ex1:eq:5 ] ) are satisfied if we choose [ ex1:eq:6 ] =( -m , a_1,a_2,a_3 ) , m = a_1 ^ 2+a_2 ^ 2+a_3 ^ 2,= ( 1,i a_1,ia_2,ia_3 ) , i^2=-1 , + _ 2=,l_2= ( ccc 0 & - ( ) & - ( ) + 0 & ( ) & - ( ) + 1 & 0 & 0 ) , + =( ) + i ( ) , = 1 , where @xmath336 , @xmath343 , @xmath340 , @xmath344 , with @xmath345 are arbitrary functions of @xmath21 and @xmath346 . the riemann invariants associated with the wave vectors @xmath335 , @xmath160 and @xmath333 take the form r_1=-m t+a_1 x+a_2 y+a_3 z , + r_2=t+i ( a_1 x+a_2 y+a_3 z ) , \|r_2=t - i ( a_1 x+a_2 y+a_3 z ) . since the quantities @xmath151 which appear in the system ( [ ex1:eq:1 ] ) are constant , the wave vectors @xmath335 , @xmath160 and @xmath333 are also constant . therefore , the partial derivatives @xmath347 all vanish , and so the matrix @xmath252 in expression ( [ eq : mm:2 ] ) is the identity , _ i.e. _ @xmath348 . consequently , the partial differential system ( [ eq : mm:5 ] ) in terms of invariants @xmath328 , @xmath329 and @xmath330 becomes [ ex : eq:8 ] = _ 1 l_1 b+_1 , [ ex : eq:9 ] = & ( 1+i m ) ( ( c t_1 + t_2 + t_3 ) + ( ) ( c b_2/b_1 + 1 + 0 ) . + & .+ ( c b_1 + b_2 + b_2 ) ) , [ ex : eq:10 ] = & ( 1-i m ) ( ( c t_1 + t_2 + t_3 ) + ( ) ( c b_2/b_1 + 1 + 0 ) . + & .+ ( c b_1 + b_2 + b_2 ) ) . note that since the derivative @xmath349 is real and the functions @xmath350 , @xmath351 and @xmath352 are real and arbitrary , they can be chosen so as to satisfy equations ( [ ex : eq:8 ] ) . instead of the system composed of ( [ ex : eq:9 ] ) and ( [ ex : eq:10 ] ) , we consider a linear combination of these equations in order to obtain an equivalent system with the real quantities @xmath353 and @xmath354 on the left hand side . the three real - valued functions @xmath223 , @xmath355 , @xmath356 appear in three of the six resulting equations . we choose these arbitrary functions in such a way that the three equations where they appear are satisfied . thus we only have to solve the three remaining equations , namely [ ex1:eq:11 ] = , = , = , + = ( 1/2)((1-i m)r_2+(1+i m)\|r_2)=t+m(a_1 x+a_2 y+a_3 z ) . if , for a given inhomogeneous term @xmath357 , the system ( [ ex1:eq:11 ] ) can be solved , then @xmath358is a solution of system ( [ ex1:eq:1 ] ) . for example , in the case where the function @xmath262 is defined by @xmath359and @xmath360 , @xmath361 , and @xmath362 , the solution of equation ( [ ex1:eq:1 ] ) takes the form [ eq : solsech ] u = a_1 ( + c_1(r_1 ) ) , v = a_2 ( + c_2(r_1 ) ) , + w = a_3 ( + c_3(r_1 ) ) , where the @xmath363 , @xmath326 , are arbitrary functions of one variable . this solution represents a solitonic bump - type wave . moreover , if the vector @xmath262 of the inhomogeneous system ( [ ex1:eq:1 ] ) is defined by b_1=-(1-k_1 ^ 2(1-u_2))^1/2(1-u^2)^1/2,b_2=-(1-k_2 ^ 2(1-v_2))^1/2(1-v^2)^1/2 , + b_3=-(1-k_3 ^ 2(1-w_2))^1/2(1-w^2)^1/2 , then the system ( [ ex1:eq:1 ] ) admits a cnoidal wave solution @xmath364another possible solution is obtained when the vector @xmath262 of the inhomogeneous system ( [ ex1:eq:1 ] ) has the form b_1=-(u^2 - 1)^1/2(u^2-k_1 ^ 2)^1/2,b_2=-(v^2 - 1)^1/2(v^2-k_2 ^ 2)^1/2 , + b_3=-(w^2 - 1)^1/2(w^2-k_3 ^ 2)^1/2 . in that case , the obtained solution is given by [ eq : solcn ] u=(1-(+c_1(r_1),k_1)^2)^-1/2,v=(1-(+c_2(r_1),k_2)^2)^-1/2 , + w=(1-(+c_3(r_1),k_3)^2)^-1/2 . equation ( [ eq : solcn ] ) represents a bounded multisolitonic solution of the system ( [ ex1:eq:1 ] ) in terms of the jacobi elliptic function @xmath365 . the moduli @xmath366 of the elliptic function are chosen in such a way that @xmath367 , @xmath326 . this ensures that the elliptic solution possesses one real and one purely imaginary period and that for the real argument we have a real - valued solution . [ [ section-17 ] ] * 2 . * + + + + consider the hydrodynamic - type system with three dependent and four independent variables [ ex2:eq:1 ] u_t+a ( u ) = b , where @xmath368 is the velocity vector , which depends on @xmath321 , @xmath369 is a constant vector and @xmath370 , @xmath371 . system ( [ ex2:eq:1 ] ) can be written in the matrix form [ ex : eq:2 ] u_t+^1 u_x+^2 u_y+^3 u _ z = b , where the matrices @xmath372 are given by @xmath373as in the previous example , we look for a mixed - type solution , _ i.e. _ of the form @xmath374where the wave vector @xmath375 is real and the wave vectors @xmath376 and @xmath377 are complex conjugates . the scalar functions @xmath250 and the special orthogonal matrices @xmath251 have to satisfy the algebraic relation ( [ ex1:eq:4 ] ) , in which @xmath336 and @xmath337 are associated with the real riemann invariant @xmath328 and @xmath338 , @xmath378 , @xmath339 , @xmath379 are associated with the complex riemann invariants @xmath329 and @xmath330 . the real vector @xmath340 associated with @xmath328 and the complex - valued vectors @xmath341 and @xmath342 associated with @xmath329 and @xmath330 have to satisfy wave relation ( [ ex1:eq:5 ] ) . the equations ( [ ex1:eq:4 ] ) and ( [ ex1:eq:5 ] ) are satisfied by the following choices @xmath380@xmath381@xmath382@xmath383@xmath384where @xmath385 is an arbitrary function of @xmath21 and @xmath15 , @xmath386 is the sign of @xmath387 and @xmath388 and @xmath389 , @xmath390 , @xmath391 , @xmath392 are arbitrary functions of @xmath15 . consequently the conditions ( [ ex1:eq:4 ] ) and ( [ ex1:eq:5 ] ) are satisfied and we obtain the reduced system [ ex2:eq:4 ] = _ 11+a_2 e^w g , = _ 12-a_1e^w g , = -+e^w , + = s_1-i ( ) + ( a_2 + i a_1 ) , + = s_2+i s_2-(a_2- i a_1 ) , + = s_3 ( 1 + ) + ( a_2+i a_1 ) , and the three conjugate equations for @xmath393 , @xmath394 and @xmath395 . [ [ section-18 ] ] one particular solution of the system ( [ ex2:eq:4 ] ) takes the form [ ex2:eq : sol ] u(r_1,r_2,\|r_2)= & ( + ) + - , + v(r_1,r_2,\|r_2)= & ( - ) + + , + w(r_1,r_2,\|r_2)=&-a_3 ^ -1 ( f_1(r_1)+a_2f_2(r_2 ) ) + & - ( -c_1-dr_1 ) , where @xmath396 is an arbitrary function of the complex riemann invariants @xmath397 and @xmath398 , @xmath399 , @xmath400 and @xmath401 are arbitrary real functions of the real invariant @xmath402 and @xmath403 , @xmath167 are real integration constants . note that , the solution is real if and only if @xmath404 holds . [ [ section-19 ] ] * 3 . * + + + + now we shall consider an example which illustrates the theoretical considerations presented in section [ sec:6 ] . we have chosen the underdetermined loewner system @xcite [ ex3:eq:1 ] u_y - v_x = h_1(u , v,),(u)_x+(v)_y = h_2(u , v , ) , where @xmath405 and @xmath406 are functions ( to be determined ) of the dependent variables , _ i.e. _ the components of the velocity of the fluid @xmath15 , @xmath407 and the density @xmath70 . system ( [ ex3:eq:1 ] ) describes a stationary compressible planar fluid flow admitting a vorticity @xmath405 and some source of matter @xmath406 . after the change of variables @xmath408the system ( [ ex3:eq:1 ] ) becomes @xmath409or equivalently in matrix form [ ex3:eq:4 ] ^1u_x+^2u_y = b , where @xmath410 , @xmath411 and [ ex3:eq:5 ] ^1= ( ccc 0 & -1 & 0 + 1 & 0 & u ) , ^2= ( ccc 1 & 0 & 0 + 0 & 1 & v ) . we are looking for a simple mode ( @xmath0=1 in ( [ eq : mmsd:1 ] ) ) , that is a @xmath412 2 solution of ( [ ex3:eq:4 ] ) of the form @xmath413 associated with the constant wave vectors @xmath414where @xmath415 is the imaginary unit . the algebraic condition ( [ eq : mm:8 ] ) , with matrices @xmath416 given by ( [ ex3:eq:5 ] ) , is satisfied for the matrix @xmath417and its complex conjugate @xmath418 , where @xmath419 and @xmath420 are arbitrary functions of independent and dependent variables . the matrix @xmath252 is equal to the identity matrix , since the wave vectors @xmath160 and @xmath333 are constant vectors . under these circumstances , the reduced system given by ( [ eq : mmsd:5 ] ) becomes [ ex3:eq:10 ] = & b_2 ( 1/2+(1+i)b_1 ) + _ 1 , , + = & b_1 ( 1/2 ) + ( 1-i)b_2+_2 , , + = & w ( b_1-b_2)(iu+v)+_3 , , where each c.c . means the complex conjugate of the previous equation . the quantities @xmath421 , @xmath422 , are the components of the characteristic vector @xmath271 which satisfies the wave relation @xmath423after solving the equation ( [ eq : mmsd:3 ] ) for the characteristic vector @xmath271 , the system ( [ ex3:eq:10 ] ) reduces to the equations [ ex3:eq:12a ] \|f_\|r - f_r = b_1(f,\|f , q ) , + f_r+\|f_\|r+fq_r+\|fq_\|r = b_2(f,\|f , q ) , where we have introduced the notation @xmath424 in particular , if we choose the right - hand side of the equation ( [ ex3:eq:4 ] ) to be @xmath425then we can solve equations ( [ ex3:eq:12a ] ) for the functions @xmath426 and @xmath427 [ ex3:eq:14 ] f=,\|f=,q=\|f\|^2 , where @xmath282 is an arbitrary function of @xmath428 and @xmath429 . by virtue of ( [ ex3:eq:14 ] ) the simple mode solution of the system ( [ ex3:eq:1 ] ) is given by [ ex3:eq:15 ] u(x , y)= & ( \|f(x- iy)-f(x+i y ) ) , + v(x , y)= & ( \|f(x- i y)+ f(x+iy ) ) , + ( x , y)=&\|f(x+i y)\|^2 . here the rank-2 solution depends on one arbitrary function @xmath282 and its complex conjugate . we have presented a variety of new approaches to solving first - order quasilinear systems in terms of riemann invariants . these methods proved to be particularly effective in delivering solutions which can be interpreted physically as superpositions between @xmath0 simple waves for hyperbolic systems or @xmath0 modes for elliptic ones . one of the interesting results of our analysis is the observation that the algebraization of these systems enables us to construct classes of solutions for which the matrix of the derivatives of unknown functions is expressible through special orthogonal matrices . this fact made it possible to obtain different types of solutions corresponding to different admissible choices of the special orthogonal matrices involved in this computation . [ [ section-20 ] ] concerning the application to the inhomogeneous fluid dynamics equations ( section [ sec : ex : gmc ] ) , we have shown that our approach is productive , leading to new interesting solutions . using the gmc we were able to construct several @xmath430 solutions in the cases of @xmath157 , @xmath158 , @xmath197 , @xmath431 and @xmath432 . each of these solutions represents a superposition of a simple wave with a simple state and involves several arbitrary functions of one variable . the first derivatives of these solutions , in most cases , tend to infinity after a finite time . it was proved @xcite that if the initial data is sufficiently small , then there exists a time interval , say @xmath433 $ ] , for which the gradient catastrophe of a solution of the system ( [ ex : gmc:1 ] ) does not occur . for these types of solutions expressible in terms of riemann invariants , we generalize the gmc in order to obtain wider classes of solutions of the inhomogeneous systems ( see sections [ sec:3]-[sec:7 ] ) . the proposed technique is applied to the inhomogeneous hydrodynamic - type systems ( [ ex1:eq:1 ] ) , ( [ ex2:eq:1 ] ) and ( [ ex3:eq:1 ] ) . we have introduced the complex integral elements instead of the real simple integral elements . this allows us to construct , based on those elements , several nonlinear superpositions of elementary solutions ( modes ) for these hydrodynamic - type systems , which to our knowledge are all new ( namely ( [ eq : solsech ] ) , ( [ eq : solcn ] ) , ( [ ex2:eq : sol ] ) and ( [ ex3:eq:15 ] ) ) . this approach proves to be an effective tool for this purpose . the proposed technique is applicable to larger classes of hyperbolic and elliptic quasilinear systems . so it is worth investigating whether our approach to construct multimode solutions can be extended to the navier stokes system in the presence of external forces ( gravitational , coriolis , etc . ) , where the dissipation effects are of particular interest . some preliminary analysis suggests that this is feasible . [ [ section-21 ] ] it is worth noting that the method for solving first - order quasilinear differential equations can be applied , with necessary modifications , to more general cases , namely to nonautonomous systems for which the coefficients depend on unknown functions @xmath15 and also on independent variables @xmath21 . an extension of our analysis to this case will be the subject of a future work . this work was supported by a research grant from the natural sciences and engineering council of canada . a mapping @xmath434 is called a simple state solution of the inhomogeneous system ( [ eq : omr:1 ] ) if all first - order derivatives of @xmath15 with respect to @xmath140 are decomposable in the following way where the real - valued functions @xmath435 and @xmath436 satisfy the algebraic relation @xmath437 in contrast to the condition @xmath438defining the simple wave solution for homogeneous systems , the expression ( [ eq : a:1 ] ) does not include a function @xmath439 of @xmath21 . consequently , the compatibility conditions are not identically satisfied and they lead to the following conditions whenever equations ( [ eq : a:1 ] ) hold , where @xmath440here we have used the notation @xmath441 . hence the system ( [ eq : a:1 ] ) has a solution if @xmath442holds . this means that the direction of @xmath128 is constant along the vector field @xmath124 . chosing a proper normalization for the wave vector @xmath443 and the vector field @xmath444 , where @xmath445 , one can obtain that @xmath83 is constant along the vector field @xmath124 , _ i.e. _ @xmath446thus the image of a solution is a curve @xmath447 tangent to the vector field @xmath124 . so one can choose a parametrization of a solution @xmath447 such that the ordinary differential equations @xmath448hold . the wave vector @xmath128 has a constant direction and by choosing a proper length of @xmath128 such that @xmath449 , we may represent a simple state solution of the inhomogeneous system ( [ eq : omr:1 ] ) in the form @xmath450where @xmath451this solution was introduced in analogy with a simple wave solution of a homogeneous system ( [ eq : omr:1 ] ) which satisfies the relations ( see _ e.g. _ @xcite ) .	we present a new technique for constructing solutions of quasilinear systems of first - order partial differential equations , in particular inhomogeneous ones . a generalization of the riemann invariants method to the case of inhomogeneous hyperbolic and elliptic systems is formulated . the algebraization of these systems enables us to construct certain classes of solutions for which the matrix of derivatives of the unknown functions is expressible in terms of special orthogonal matrices . these solutions can be interpreted as nonlinear superpositions of @xmath0 waves ( or @xmath0 modes ) in the case of hyperbolic ( or elliptic ) systems , respectively . theoretical considerations are illustrated by several examples of inhomogeneous hydrodynamic - type equations which allow us to construct solitonlike solutions ( bump and kinks ) and multiwave ( mode ) solutions .
the hottest helium - rich white dwarfs exhibit lines of ionized helium in their optical spectra and they are classified as do white dwarfs . currently forty dos are known . the coolest one has @xmath1 k and the hottest one is kpd0005 + 5106 . from an analysis of optical and _ hubble space telescope _ fos spectra @xmath2 k and @xmath3 has been derived ( werner et al . 1994 ) for kpd0005 + 5106 . the sloan digital sky survey has recently revealed another do with @xmath4 k ( hgelmeyer et al . 2006 ) . in the _ far ultraviolet spectroscopic explorer _ ( fuse ) spectra of the hottest pg1159 stars ( @xmath5 k ) we have recently discovered absorption lines of ( werner et al . the same features were also discovered in kpd0005 + 5106 ( fig.1 ) . emission lines in the optical spectrum are also from and not , as previously thought , from superionized ( i.e. , non - thermally excited ) . non - lte line profile fits to the lines of kpd0005 + 5106 yield @xmath0 k and @xmath6 . a reassessment of the line spectrum confirms these parameters ( werner et al . hence , kpd0005 + 5106 is by far the hottest he - rich white dwarf . in fact , it is the hotter than any other white dwarf or central star of planetary nebula . it is only rivaled by the exotic pg1159 star h1504 + 65 ( werner et al . 2004 , and these proceedings ) . rosat has revealed two x - ray emission components of kpd0005 + 5106 . _ chandra _ observations of the soft component ( 2080 ) prove that it is of photospheric origin ( drake & werner 2005 ) and not , as claimed earlier , coronal ( fleming et al . deposing the idea of non - photospheric lines is in accordance with the deposition of the corona . the hard component ( at 12 ) remained unexplained ( odwyer et al . 2003 ) . in the light of the newly determined , extreme @xmath7 it should be investigated whether this component is of photospheric origin , too . kpd0005 + 5106 is located on a 0.7m@xmath8 wood & faulkner ( 1986 ) post - agb track , just before the `` knee '' , at the hot end of the white dwarf cooling sequence . this position is well within the pg1159-class domain of the hr diagram , hence , it can not simply be a pg1159 descendant . the immediate progenitor could have been a o(he ) star . this spectral class consists of five hot stars ( @xmath9 k , @xmath10 , rauch et al . 1998 , and these proceedings ) with almost pure he atmospheres . kpd0005 + 5106 as well as the o(he ) stars might be descendants of rcrb stars . is supported by the _ german astrophysical virtual observatory _ project of the german federal ministry of education and research under grant 05ac6vtb . is supported by the fuse project , funded by nasa contract nas5 - 32985 . drake , j. j. , & werner , k. 2005 , apj , 625 , 973 fleming , t. a. , werner , k. , & barstow , m. a. 1993 , apj , 416 , l79 hgelmeyer , s. d. , dreizler , s. , homeier , d. , et al . 2006 , a&a , 454 , 617 odwyer , i. j. , chu , y .- h . , gruendl , r. a. , et al . 2003 , aj , 125 , 2239 rauch , t. , dreizler , s. , & wolff , b. 1998 , a&a , 338 , 651 werner , k. , heber , u. , & fleming , t. a. 1994 , a&a , 284 , 907 werner , k. , rauch , t. , barstow , m. a. , & kruk , j. w. 2004 , a&a , 421 , 1169 werner , k. , rauch , t. , & kruk , j. w. 2007 , a&a , in press wood , p. r. , & faulkner , d. j. 1986 , apj , 307 , 659	kpd0005 + 5106 is the hottest known helium - rich white dwarf . we have identified lines in uv and optical spectra and conclude that it is significantly hotter than previously thought , namely @xmath0 k instead of 120000 k. this is a possible explanation for the observed hard x - ray emission as being of photospheric origin . concerning its evolutionary state , we suggest that kpd0005 + 5106 is not a descendant of a pg1159 star but more probably related to the o(he ) stars and rcrb stars .
apart from their evident importance in experimental tests of the foundations of quantum mechanics , entangled photons promise many applications to quantum information processing @xcite , secure quantum communication @xcite , lithography @xcite or metrology @xcite . in addition , their non - classical frequency and time correlations could also open up novel spectroscopic applications @xcite . entanglement - induced two - photon transparency @xcite and the linear scaling of two - photon induced fluorescence with the pump intensity @xcite constitute two basic non - classical effects observed with entangled photon pairs . this scaling makes it possible to carry out nonlinear optical measurements with much lower light intensity compared to classical light , as we will discuss in section [ lightsources ] . more generally , entangled photons offer new control parameters for nonlinear spectroscopy @xcite , and can be used to distinguish quantum pathways of matter @xcite . it remains however an open question , to what extent quantum entanglement is essential for these effects and whether some can be reproduced , for instance , by shaped or stochastic classical pulses @xcite . recently , it has been argued that entanglement can induce collective resonances between non - interacting two - level atoms @xcite . it was later found that these resonances may not be observed in fluorescence due to destructive interference of pathways @xcite . + , at the same time the manifold @xmath0 is well - resolved due to the narrow distribution of @xmath1 . b : transitions with classical light . bandwidths of @xmath2 and @xmath3 now add up , and the f - manifold can not be resolved.,scaledwidth=45.0% ] in this paper we present a theoretical framework for describing excitations induced by entangled photons in chromophore aggregates . these possess manifolds of single- and double - exciton states @xcite , and the excitation of each double - exciton state can take place via many interfering pathways ( see figure [ transitions ] ) . we show that the non - classical spectral and temporal properties of twin photons can be used to excite state distributions that may not be obtained by classical light . due to energy conservation in their generation , the sum of the energies of twin photons @xmath4 has a sharp distribution ( @xmath5 @xmath6 is very small ) , even though each single photon may be broadband ( @xmath7 very large ) . the high frequency resolution of @xmath8 allows to target certain double - exciton states , while the broad bandwidth of the individual beams enables the simultaneous access of all possible pathways that reach the double - exciton state ( see figure [ transitions ] ) . in contrast , with classical fields the bandwidths of the two beams add up , and create a large uncertainty in the sum . our simulations show that the non - classical distributions could be detected by the fluorescence signal . the control parameters of the entangled light can be used to stretch the signal in two - dimensional plots , that could reveal additional information about the matter . we consider a chromophore aggregate interacting with the electromagnetic field and described by the hamiltonian @xmath9 @xmath10 , @xmath11 and @xmath12 represent the aggregate , the field and their coupling , respectively . the electronic states group into well - separated manifolds , which are denoted as e- ( single - exciton ) , f- ( double - exciton ) manifold and so on . the field couples to the system via the dipole operator , which induces transitions between these manifolds . in the rotating wave approximation , the interaction hamiltonian reads @xmath13 where we have introduced the positive - frequency component of the dipole operator @xmath14 in which @xmath15 are matter transition frequencies , and @xmath16 the dipole moments . the corresponding negative - frequency part of the electromagnetic field operator is given by @xmath17 where @xmath18 denotes the quantization volume , @xmath19 is the creation operator for mode s , and s runs over the relevant modes . assuming perfect phase matching , we can neglect the spatial dependencies of the electromagnetic field . this is justified for large samples . coupling with a phonon bath can induce many interesting relaxation effects . these are not included in the present study . we first derive general expressions for density matrix elements in the single- and the double - exciton manifold in terms of convolutions of matter- and field - correlation functions . these results are valid for light with arbitrary statistical properties , and can describe two - photon induced fluorescence and related measurements in a consistent manner . the matrix element @xmath20 of the single - exciton part of the density matrix is given by @xmath21 , \label{pe(t)}\ ] ] where @xmath22 denotes the density matrix of the entire matter - field system . in a superoperator liouville space representation , the formal solution of the heisenberg equation is given by the dyson series @xcite @xmath23 where @xmath24 is the time - ordering operator , and the superoperator @xmath25 is defined through @xmath26 in the interaction picture with respect to @xmath27 . for weak fields we can expand the exponential , and the leading ( second- ) order contribution to eq . ( [ pe(t ) ] ) yields @xmath28 where @xmath29 , and @xmath30 denotes the set of control parameters that govern the excitation , @xmath31 frequencies , pulse envelopes etc . the square brackets @xmath32 denote the quantum mechanical expectation value with respect to the initial state @xmath33 , @xmath34 @xmath5 the aggregate is in the ground state @xmath35 , and the electromagnetic field is prepared in a state @xmath36 . + the fourth - order processes that end in the @xmath37 manifold can be represented by the diagrams i and ii shown in figure [ single2 ] ( for diagram rules , see @xcite ) . these can be divided into two classes : raman type that only involve single - exciton resonances ( diagram i ) , and two - photon absorption type ( diagram ii ) that involve the double - exciton manifold . we have to take two additional diagrams into account , in which three interactions occur on the right - hand side . apart from the projector @xmath38 , these are simply the complex conjugates of diagrams i and ii in figure [ single2 ] . we can thus write them as @xmath39 note that for diagonal density matrix elements , @xmath5 @xmath40 , we can combine the two terms in both ( [ pei ] ) and ( [ peii ] ) , and recast them as the real part of the first term . to the same order , double - exciton matrix elements are given by diagram iii of figure [ single2 ] , @xmath41 where @xmath42 . ( [ pei ] ) , ( [ peii ] ) and ( [ pf ] ) enable us to evaluate the density matrix for arbitrary states of the radiation field once we specify its four - point field correlation function . creates two beams of entangled photons with central frequencies @xmath2 and @xmath3 , such that @xmath4 . the central frequencies are determined by the relative angles between the direction of propagation.,scaledwidth=40.0% ] extensive research effort has focused on producing and detecting pairs of entangled photons . numerous schemes have been employed for the generation of entangled photon pairs , including parametric down - conversion @xcite , biexciton decay in semiconductors @xcite or four wave mixing in optical fibers @xcite . + we consider a type of entangled light known as twin photons @xcite . it is created by type - ii parametric down - conversion with a cw - pump laser of frequency @xmath8 . when the pump beam is sufficiently weak , it predominantly produces temporally non - overlapping pairs of entangled photons . the state of the field can then be obtained pertubatively in the interaction hamiltonian of the pump beam with the nonlinear crystal @xcite @xmath43 where @xmath44 is the pump field amplitude . the state is not normalized . here , the length of the crystal along the pump beam is denoted @xmath45 , while it is assumed to be infinitely large in the other two directions . in the setup depicted in figure [ setup ] , one collects the light in two outgoing directions @xmath46 and @xmath47 with respect to the optical axis , thus fixing the central frequencies @xmath2 and @xmath3 of the photon wavepackets in directions 1 and 2 . the photon pairs are furthermore correlated in time . this is characterized by the entanglement time @xmath48 , which is determined by the length @xmath45 along the axis of the pump beam . therefore , one can vary independently @xmath48 and two of the three frequencies @xmath49 and @xmath3 . + for short @xmath48 , eq . ( [ entangledstate ] ) leads to the broad power spectrum @xmath50 + \text{sinc}^2 \big [ \frac{(\omega - \omega_2)t}{2 } \big ] \right ) , \label{ntwin}\end{aligned}\ ] ] which is depicted in figure [ powerspectrum ] . ) ] with @xmath51 . the central frequencies are @xmath52 and @xmath53 . the two sinc@xmath54-functions overlap to form a single broad peak . ] to evaluate the normally ordered correlation functions in eq . ( [ peii ] ) , we note that it can be separated into @xmath55 where @xcite @xmath56 here , we had defined the rectangular function @xmath57 the non - classicality of the quantum state produced by this setup has been verified in numerous experiments @xcite . one important feature that follows from eqs . ( [ factorization ] ) and ( [ fourthorder ] ) is that the four - point correlation function of the fields scales linearly with the pump intensity @xmath58 , while classically it scales quadratically @xcite . this scaling behavior can be utilized in experiments to determine intensity regimes , in which eq . ( [ entangledstate ] ) can account for the produced signal . in practice , we expect this linear scaling of fourth - order signals with a crossover to a quadratic classical scaling at higher intensities . the light beam can be considered to be made of pairs of photons . at weak intensities the pairs are temporally well separated , and the process is induced by two photons of the same pair . at higher intensities it becomes statistically more plausible for the two photons to come from different pairs , which are not entangled and the classical scaling is recovered . this crossover has been demonstrated experimentally @xcite . for instance , in @xcite the authors detect it at an intensity ( of the entangled photon flux ) of @xmath59 in a ppktp crystal , which was obtained by a pump intensity of less than @xmath60 . the linear scaling of the two - photon absorption was also demonstrated experimentally for organic porphyrin dendrimers @xcite . + entangled photon sources are weak , but at the same time weak intensities are required to see the effect of entanglement . nonlinear spectroscopy and imaging with weak fields should allow to limit the damage to biological samples . to highlight the effects of entanglement , we consider a classical reference state of the field with the same power spectrum . to that end , we introduce the frequency decomposition of a classical field @xmath61 where @xmath62 and @xmath63 are real functions representing the amplitude and the phase of the various modes . to reproduce the power spectrum eq . ( [ ntwin ] ) , we took @xmath64 e^{- i \phi(\omega ) } , \label{a_1}\\ a_2 ( \omega ) & = a_0 \ ; \text{sinc } \big[(\omega - \omega_2 ) t/ 2 \big ] e^{- i ( \phi(\omega ) + \pi / 2)}. \label{a_2}\end{aligned}\ ] ] with this choice @xmath65 does not affect the power spectrum and will be chosen to obtain stationary stochastic light ( rather than a pulse ) . we assume that @xmath65 is a random function , so there is no well - defined phase relation between the different frequencies . the quantum mechanical expectation value in correlation functions such as @xmath66 then needs to be replaced by an average over the distribution of @xmath65 : @xmath67 assuming that @xmath68 is uniformly distributed in the interval @xmath69 we find that @xmath70 is a stationary gaussian process @xmath71 classical light only shows correlations on the level of intensities , @xmath5 @xmath72 . by substitution of eqs . ( [ a_1 ] ) and ( [ a_2 ] ) we then obtain @xmath73 , \label{classicalcorr}\end{aligned}\ ] ] where the summations run over @xmath2 and @xmath3 , and tri@xmath74 is the triangular function as defined in ( [ triangular ] ) . this correlation function represents stationary , stochastic light , since it only depends on time differences , just like the one produced by entangled photons with stationary pump [ see eq.([fourthorder ] ) ] . however , there is one crucial difference . the quantum mechanical expectation value can be factorized in eq . ( [ factorization ] ) , and only depends on @xmath75 and @xmath76 . the correlation function ( [ classicalcorr ] ) depends on @xmath77 , @xmath78 , @xmath79 and @xmath80 . we now present simulations of the exciton density matrices produced by entangled and stochastic light with the same spectral density in a model of the bacterial reaction center ( rc ) of purple bacterium b. viridis @xcite . the primary steps in photosynthesis involve excitation energy transfer towards and charge separation within rcs . the rc of purple bacteria utilizes the high intensity near - ir region of the solar irradiation , and transforms the energy into a chemical potential gradient with near unity efficiency . in bacterial rc the charge separation occurs in the active branch of the protein on the sub - ps to low ps timescale @xcite . the quenching of the excited state special pair population has been demonstrated experimentally in time resolved emission measurements @xcite . + the electronic hamiltonian @xmath10 describes the optically bright chromophore excitations and dark charge separated states in the active branch of the rc in a tight - binding formulation . it includes 12 single- and 41 double - exciton states as depicted in figure [ spectrum]c . to construct @xmath10 , we start with the x - ray structural data for b. viridis ( pdb code : 1prc ) @xcite . the q@xmath81 transition dipole moments are placed at the center of bacteriochlorophylls and bacteriopheophytines . the excitation energies of bcl s and bp s are taken from ref . @xcite , frster couplings are calculated in the dipole approximation , except for the special pair where a value of 852 @xmath82 was chosen , allowing to reproduce the cd spectra of b. viridis . the energy of the primary charge separated ( cs ) state is fixed relative to the lowest special pair exciton state @xcite , which yields a reference point for the additional cs energies of the active branch . the hamiltonian is given in detail in ref . @xcite.dephasing effects and incoherent population transport were neglected . to model the absorption spectrum shown in figure [ spectrum ] , we assume the same lifetime broadening @xmath83 for all single - exciton states , which agrees reasonably with more detailed simulations @xcite . in figure [ spectrum]b , we mark the single - exciton states @xmath84 by vertical lines , whose length is proportional to their oscillator strength @xmath85 . the spectrum is dominated by six states corresponding to molecular frenkel excitations of the constituents of the rc . the other six charge transfer states are dark , and can not be accessed spectroscopically . + the density matrix induced by entangled or stochastic light is block - diagonal in the e- and f - manifolds , since a coherence between the two manifolds can be related to field correlation functions of the kind @xmath86 or @xmath87 , which vanish for any fock state or stationary gaussian process . in figure [ densitymatrix ] we depict the absolute values of the density matrix elements of the double - exciton states @xmath88 to @xmath89 for @xmath90 and @xmath91 . entangled light excites a pure double - exciton state , @xmath92 [ see eq . ( [ tfgentangled ] ) ] . consequently the purity of the density matrices in figure [ densitymatrix]a - b is @xmath93 . the density matrix induced by stochastic light ( figure [ densitymatrix]c - d ) is not in a pure state , we obtain @xmath94 independent of the pump frequency . furthermore , it is apparent that by tuning the pump frequency , we can select the excitation of certain states . figure [ densitymatrix]a shows strong excitation of the states @xmath95 , @xmath96 and @xmath97 , while in figure [ densitymatrix]b the states between @xmath98 and @xmath89 are most strongly excited . this selectivity allows to manipulate of the fluorescence signal , as will be shown in the next section . while the density matrices in figures [ densitymatrix ] a & b strongly depend the pump frequency , our calculations of the corresponding density matrices produced by stochastic light indicate no such selectivity . the results are plotted in figures [ densitymatrix ] c & d. for both pump frequencies ( @xmath99 and @xmath100 @xmath82 ) , the excitation is distributed among the bright states @xmath101 and @xmath89 . + figure [ densitymatrixe ] shows the single - exciton manifold density matrices [ eq . ( [ peiientangled ] ) ] for the same pump frequencies @xmath102 and @xmath100 @xmath82 . our results qualitatively resemble the ones for the double - exciton manifold . by varying @xmath8 we can enhance ( figure [ densitymatrixe]a ) or suppress ( figure [ densitymatrixe]b ) the excitation of @xmath103 ( and , vice versa , of @xmath104 and @xmath105 ) , while stochastic light always creates very similar density matrices . for both frequencies , the population is mainly distributed between the states @xmath103 and @xmath105 . , eq . ( [ s2a ] ) , with @xmath106 . ] ( [ pf ] ) ] plotted vs. @xmath3 and @xmath8 . a : excitation by entangled light with @xmath106 . b : excitation by stochastic light . ] the total population of the single - exciton manifold @xmath107 calculated to second order in the matter - field interaction [ eq . ( [ peentangled ] ) ] is displayed in figure [ pe2d ] vs. the frequencies @xmath8 and @xmath3 . due to the broadband nature of the beams , none of the peaks of the absorption spectrum in figure [ spectrum]b can be resolved . this population only depends on the spectral density , is identical for stochastic and entangled light , and is not susceptible for the non - classical features of entangled light . + the total double - exciton population @xmath108 created by eq . ( [ tfgentangled ] ) is plotted in figure [ manifolds]a . the distribution of the single - exciton states to fourth order ( see diagram ii in figure [ single2 ] ) looks almost identical , since eq . ( [ peii ] ) includes an excitation to the double - exciton manifold , and is not shown . it depends only weakly on @xmath3 , reflecting the energy uncertainty due to the time entanglement . thus , the central frequency of each beam is not important , and in the following we set @xmath109 . two broad resonance bands , at @xmath110 ( band @xmath111 ) , and at @xmath112 ( band @xmath113 ) , dominate the distribution . from our previous discussion of the density matrix , we anticipate that states @xmath114 and @xmath97 dominate band @xmath111 , whereas the states @xmath115 seem to make up band @xmath113 ( see figure [ densitymatrix ] ) . the signal induced by stochastic light ( figure [ manifolds]b ) shows no such structure . due to the broad bandwidth , the level structure of the rc can not be resolved at all . + with @xmath106 upon excitation by entangled light . insets : excitation by stochastic light . b : same for the single - exciton states ] to get a more detailed picture of this state distribution , we depict the contribution of various double - exciton state populations to the distribution [ eq . ( [ tfgentangled ] ) ] in figure [ pfsingle]a . the distributions are normalized at each @xmath8 . states @xmath95 , @xmath96 and @xmath97 dominate band @xmath111 , whereas band @xmath113 is dominated by states @xmath116 , @xmath117 and @xmath118 . at higher pump frequency , states @xmath98 , @xmath89 and @xmath119 are most pronounced . in general , the regions of leading contribution group around the states energies , but the distributions may be asymmetric because of the presence of other states . we next turn to the distribution of single - exciton states ( figure [ pfsingle]b ) . since the single excitons are obtained by emission from a two - exciton state , the distribution closely resembles the double - exciton distribution in figure [ pfsingle]a . for instance , when @xmath120 , the double - exciton state @xmath95 is on resonance . it decays primarily into single - exciton state @xmath103 , and accordingly the contribution of this state increases around the same value of the pump frequency . the same holds for , e.g. , the double - exciton state @xmath89 , and the single - exciton state @xmath121 . however , in most cases the different contributions overlap , and the relation between the single - exciton and the double - exciton states is more complex . having discussed our simulations of the excited state distribution created by entangled light , we now turn to the question whether it may be possible to detect fingerprints of these distributions in experiments . to that end , we calculate the dispersed fluorescence signal . this involves transitions between double- and single - exciton states as well as transitions between single - exciton states and the ground state , @xmath5 @xmath122 here , @xmath123 is the emitted fluorescence frequency , and @xmath124 ( @xmath125 ) the population of state f ( e ) given by eqs . ( [ pf ] ) and ( [ peii ] ) , respectively . more elaborate detection ( gated time and frequency , photon statistics ) is possible @xcite , but will not be considered here . + in figure [ 2d3fs ] , we display this signal vs. the pump frequency @xmath8 and the emission frequency @xmath123 . the delta functions in ( [ newsignal ] ) and ( [ signal2 ] ) were replaced by a lorentzian , which means the width of the peaks in horizontal direction of figure [ 2d3fs ] is instrumental . in analogy to the populations shown in figure [ manifolds ] , the fluorescence simulation with entangled photons ( [ 2d3fs]a ) shows two distinct resonances along the @xmath8-axis pertaining to bands @xmath111 and @xmath113 , respectively ( see also figure [ wsintegrated ] ) . along @xmath123 ( horizontal axis ) , it contains two contributions , one around @xmath126 , and a higher - energy part between @xmath127 and @xmath128 , reminiscent of the absorption spectrum . this pronounced structure can be exploited to enhance or suppress certain features , as will be demonstrated in the following . the simulation of the stochastic signal in figure [ 2d3fs]b shows no structure along the @xmath8-axis , and consequently does not allow a manipulation of the fluorescence signal . and @xmath129 , b : enlarged region by the black box . c : enlarged region for @xmath130 . d & e : same as a & b , but for stochastic light.,scaledwidth=47.0% ] the dispersed fluorescence simulation for @xmath129 is shown in figure [ wp22160]a . most of the peaks can be directly attributed to the energy of single - exciton states , the transition thus correspond to either fluorescence from the single - exciton manifold , or from double - exciton states , whose energies closely match the sum of two single - exciton states . however , the simulations also show side peaks of the resonance at @xmath131 , which can not be assigned to single - exciton energies . the strongest of these peaks is highlighted by the black arrow in the inset figure [ wp22160]b . it corresponds to the @xmath132 transition . since state @xmath95 is most strongly excited at the given pump frequency ( see figure [ pfsingle]a ) , this peak is most pronounced . at pump frequency @xmath133 resonantly excited states @xmath96 and @xmath97 show different transitions into the single - exciton state @xmath104 ( see figure [ wp22160]c ) . the simulation with stochastic light shown in figures [ wp22160]d and e can not resolve any of these sidepeaks due to its broadband nature . ) vs. the pump frequency @xmath8 : signal created by entangled photons ( solid , blue line ) , the stochastic light as ( dashed , gray line ) , and signal created by two monochromatic beams at @xmath2 and @xmath3 ( dotted , red line).the double - exciton energies are plotted as solid , blue lines , and their heights are given by @xmath134 . the single - exciton energies are indicated as dashed , red lines , with the height proportional to @xmath135 . ] we had derived expressions for the single- and double - exciton density matrices of quantum systems interacting with arbitrary light sources perturbatively in the field . these were used to simulate the excitation of matter via entangled twin photons , and stochastic light with the same power spectrum . applications to the reaction center of purple bacteria show that the populations strongly depend on the nature of the light , and we indicated how these properties could be observed in the frequency - resolved fluorescence measurements . the non - classical spectral profile of entangled light allows us to target specific double - exciton states , and to explore all the excitation pathways to this state in a single shot . this shows up perspiciously in the fluorescence action spectrum in figure [ wsintegrated ] . the spectrum reveals the level structure of the double - exciton manifold , whereas the stochastic light can not resolve this structure . for completeness , we also plot the action spectrum created by two cw laser beams with frequencies @xmath2 and @xmath3 at the peaks of the spectral density . even though the narrow bandwidth of these beams allows for a good @xmath8-resolution , this signal is most pronounced , when one of the laser beams is resonant with a single - exciton state , and thus closely resembles the linear spectrum in figure [ spectrum ] . + the role of entanglement and coherent transport in the efficiency of exciton transport in photosynthetic complexes has been subject of intense debate @xcite . using our formalism , one could also discuss the entanglement of quasiparticles in the double - exciton manifold @xcite . this will require to decompose double - exciton states into single excitons , which goes beyond the scope of the present paper . + other possible extensions are to excitation by pulsed light , and to include time- and frequency gating in order to obtain control over the time and frequency resolution of the detection . the time resolution could allow us to monitor the dynamics of exciton transport and charge separation after interaction with entangled light in an experimentally feasible way . we gratefully acknowledge the support of the national science foundation through grant no . che-1058791 , the chemical sciences , geosciences and biosciences division , office of basic energy sciences , office of science , us department of energy , and the defense advanced research projects agency ( darpa ) , grant uoft-49606 . b. p. f. gratefully acknowledges support from the alexander - von - humboldt foundation through the feodor - lynen program . using eq . ( [ entangledstate ] ) , the field correlation function of the leading - order contribution , given by eq . ( [ s2a ] ) , yields @xmath136 where we have defined the triangular function @xmath137 expanding the matter correlation function into a sum - over - states expression , we obtain @xmath138 where we dropped the constant factor c. the matrix element of the dipole operator connecting the states g and e is denoted @xmath139 , the energy difference is given by @xmath140 , and the lifetime broadening is denoted by @xmath141 . we can assume that @xmath142 . since the entanglement time @xmath48 is short with relative to other timescales in the system ( @xmath143 ) , this contribution to the single - exciton states depends weakly on the pump frequency @xmath8 . + the field correlation function of eq.([pei ] ) can be recast as a normally ordered term plus a commutator term , which we neglect . ( [ fourthorder ] ) and dropping the constant factor @xmath144 , we obtain for the populations @xmath145 this contribution is proportional to the entanglement time @xmath48 , reflecting the fact that in ( [ pei ] ) the time pairs ( @xmath146 , @xmath147 ) as well as ( @xmath148 , @xmath149 ) are correlated . this means that the entire process has to happen within the entanglement time @xmath48 . since this time is very short with respect to other timescales in the system , @xmath150 can be neglected . the only fourth - order contribution to the single - exciton manifold is thus given by ( [ peii ] ) . in a sum - over - states expression , it reads @xmath151 , \label{peiientangled}\end{aligned}\ ] ] where the summations a , b , c and d runs over @xmath2 and @xmath3 . it is apparent that the single - exciton peaks are multiplied by phase factors of the kind @xmath152 , while the two - exciton peak is not . additionally , taking into account @xmath153 the single - exciton peaks have a very similar structure as ( [ peentangled ] ) , and depend weakly on @xmath8 . thus , @xmath154 is dominated by the lorentzian @xmath155 , and this in turn means that spectroscopy with entangled photons can be used to probe the two - exciton manifold of aggregates . + we further note that eq . ( [ pf ] ) can be recast as the product of transition amplitudes @xmath156 given by @xmath157 clearly , the single - exciton resonances are suppressed , and the detuning between @xmath8 and the two - exciton manifold completely determines @xmath158 . due to the different intensity scaling of classical signals , the main contribution to the signal will arise from the leading - order contribution in eq . ( [ s2a ] ) . since the stochastic light reproduces the spectral density of entangled light , we obtain the same result as for entangled light [ see eq . ( [ peentangled ] ) ] . additionally , one can filter out this contribution , and calculate ( [ peii ] ) and ( [ pf ] ) to study the electronic excitations induced by higher - order interaction with stochastic light . + using eqs . ( [ pei ] ) , ( [ pf ] ) and ( [ classicalcorr ] ) , the fourth - order contributions to the single - exciton manifold yield where we neglected the small imaginary part of the sinc@xmath54-functions . the summations a , b run over @xmath2 and @xmath3 , and we can assume that @xmath160 . in comparison with the excitations induced by entangled photons , we observe that the single - exciton peaks in ( [ peiientangled ] ) and ( [ tfgentangled ] ) have a very similar structure to ( [ peiistochastic ] ) and ( [ pfstochastic ] ) , respectively . however , eqs . ( [ peii ] ) and ( [ pf ] ) also contain lorentzian factors that depend on @xmath1 , which dominate the signal . those resonances are replaced by a constant background contribution in the second lines of eqs . ( [ peiistochastic ] ) and ( [ pfstochastic ] ) .	the non - classical spectral and temporal features of entangled photons offer new possibilities to investigate the interactions of excitons in photosynthetic complexes , and to target the excitation of specific states . simulations of fluorescence in the bacterial reaction center induced by entangled light demonstrate a degree of selectivity of double - exciton states which is not possible using classical stochastic light with the same power spectrum .
the mbius geometry of space curves was mainly developed in the first half of the past century @xcite and later taken up starting from the early 1980 s @xcite . further developments of the subject as an instance of the conformal geometry of submanifolds can be found in @xcite and the literature therein . the subject has also received much attention for its many fields of application , including the theory of integrable systems @xcite , the topology and mbius energy of knots @xcite , and the geometric approach to shape analysis and medical imaging @xcite . 0.1 cm let @xmath1 , @xmath2 , be a smooth curve parametrized by arclength @xmath3 . the _ conformal arclength _ parameter @xmath4 of @xmath5 is defined ( up to a constant ) by @xmath6 where @xmath7 is the standard scalar product on @xmath8 . the 1-form @xmath9 , the _ infinitesimal conformal arclength _ of @xmath5 , is conformally invariant . if @xmath10 , for each @xmath3 , the curve is called _ generic_. topology ( cf . @xcite ) . ] the conformal arclength @xmath4 gives a conformally invariant parametrization of a generic curve . we consider the conformally invariant variational problem on generic curves defined by the _ conformal arclength functional _ @xmath11 = \int_\gamma \eta_\gamma.\ ] ] this variational problem was studied in @xcite for @xmath12 and more recently in @xcite for higher dimensions . in the literature . however , observe that @xmath13 has the dimension @xmath14 , so that @xmath9 is dimensionless . ] accordingly , the critical curves can be found by quadratures and explicit parametrizations are given in terms of elliptic functions and integrals . in this paper we address the question of existence and properties of closed critical curves for the functional @xmath15 . actually , it suffices to consider the 3-dimensional case only , since from the results in @xcite we can see that any closed critical curve in @xmath8 lies in some @xmath16 , up to a conformal transformation . it is known that a generic space curve is determined , up to conformal transformations , by the conformal arclength and two conformal curvatures ( cf . section [ s:1 ] ) . as for a closed critical curve with constant conformal curvatures , one can see that it is conformally equivalent to a closed rhumb line ( loxodrome ) of a torus of revolution ( cf . example [ e:1 ] ) . 0.1 cm the purpose of this paper is to study the class of closed critical curves with non - constant conformal curvatures , for brevity called _ conformal strings_. we begin by describing our three main results . if @xmath5 , @xmath17 are two curves and @xmath18 $ ] , @xmath19 $ ] denote their trajectories , then @xmath5 and @xmath20 are said _ mbius _ ( _ conformally _ ) _ equivalent _ if there is an element @xmath21 of the mbius group @xmath22 of @xmath23 , such that @xmath24=[\widetilde{\gamma}]$ ] . by a _ conformal symmetry _ of a curve @xmath5 is meant an element @xmath25 , such that @xmath24=[\gamma]$ ] . the set of all symmetries of @xmath5 is a subgroup @xmath26 of @xmath22 . the symmetry group of a closed curve with non - constant conformal curvatures is finite and its cardinality is called the _ symmetry index _ of @xmath5 . our first main result is the following . [ thm : a ] the mbius classes of conformal strings are in 1 - 1 correspondence with the rational points of the complex domain @xmath27 the rational points of @xmath28 are called the _ moduli _ of conformal strings . using this theorem and other technical results , we will prove that any mbius class of strings is represented by a model string . this is our second main result . [ thm : b ] the conformal strings corresponding to a modulus @xmath29 are mbius equivalent to a model string @xmath30 , @xmath31 called the _ symmetrical configuration _ of @xmath32 . here @xmath33 where @xmath34 , @xmath35 , @xmath36 and where @xmath37 and @xmath38 are real parameters , uniquely defined by @xmath32 , such that @xmath39 , @xmath40 , @xmath41 , and @xmath42 . the symmetry group of a symmetrical configuration has a special structure , which is described by the following . [ thm : c ] let @xmath43 be the symmetrical configuration corresponding to the modulus @xmath44 , where @xmath45 , @xmath46 , and the pairs @xmath47 , @xmath48 are coprime integers . let @xmath49 be the least common multiple of @xmath50 and @xmath51 , and consider the coprime integers @xmath52 and @xmath53 . then , 1 . @xmath49 is the order of the symmetry group of @xmath54 ; 2 . @xmath55 and @xmath56 are the linking numbers of @xmath54 with the clifford circle @xmath57 and the @xmath58-axis , respectively . an important consequence of the previous results is that the conformal shape of a string is uniquely determined by three phenomenological invariants : the order of its symmetry group and its linking numbers with the two axes of the symmetry group . the explicit construction of a string from the phenomenological invariants requires the inversion of the period map of @xmath15 ( cf . section [ s:2 ] ) . in this respect , some numerical experiments carried out with the software _ mathematica _ suggest that conformal strings are simple curves ( cf . section [ s:6 ] ) . however , a rigorous proof of this fact is still missing . another interesting problem is to find an estimate for the asymptotic growth of @xmath59 , the cardinality of the set of mbius classes of strings with symmetry order @xmath49 . numerical experiments suggest a quadratic growth of @xmath59 . 0.1 cm the theorems above have a conceptual explanation within the general scheme of griffiths formalism of the calculus of variations @xcite . using griffiths formalism , the _ momentum space _ of the variational problem can be identified with @xmath60 , where @xmath61 is the identity component of the mbius group of @xmath62 and @xmath63 is a 3-dimensional submanifold of @xmath64 , the dual of the lie algebra of @xmath61 . moreover , the restriction @xmath65 of the liouville form of @xmath66 defines an invariant contact structure on @xmath67 . by choosing a suitable set of coordinates on @xmath63 , say @xmath68 , @xmath69 and @xmath70 , the characteristic curves of the contact form are given by @xmath71 where @xmath72 is the canonical lift of a stationary curve @xmath5 , @xmath68 and @xmath69 are the conformal curvatures , and @xmath73 . from a theoretical point of view , the study of the variational problem is equivalent to that of the dynamical system defined by the characteristic vector field of the contact form @xmath74 . one can easily see that the contact momentum map is given by @xmath75 ( for the construction of the momentum map in contact geometry , see for instance @xcite . ) moreover , the action of @xmath61 on @xmath67 is hamiltonian and coisotropic , and the characteristic vector field is _ collective completely integrable _ @xcite . consequently , the flow can be linearized on the fibers of the momentum map and its trajectories can be found by quadratures ( see @xcite for a general description of the integration procedure ) . theorems [ thm : a ] and [ thm : c ] say that the contact dynamical system is quantizable , at least in the sense of `` the old quantum theory '' ( bohr s atom theory ) @xcite , and that the quantum numbers of the closed ( quantizable ) trajectories have a precise geometric meaning . 0.1 cm the paper is organized as follows . section [ s:1 ] collects some basic facts about the conformal geometry of space curves . section [ s:2 ] recalls the euler lagrange equations of the variational problem and discusses the example of closed critical curves with constant conformal curvatures . then , after introducing the natural parameters of a critical curve with non - constant periodic conformal curvatures , the period map of the functional @xmath15 is defined , and conformal strings are characterized in terms of the natural parameters , via the period map . section [ s:3 ] deals with a technical result about the period map , from which theorem [ thm : a ] follows directly . section [ s:4 ] proves theorem [ thm : b ] , while section [ s:5 ] proves theorem [ thm : c ] . section [ s:6 ] discusses some examples . 0.1 cm numerical and symbolic computations , as well as graphics , are made with the software _ mathematica_. as basic references for the theory of elliptic functions and integrals we use the monographs @xcite . for the few notions of knot theory used in the paper we refer to @xcite . a general reference for mbius geometry is @xcite , to which we refer for an updated list of modern and classical references to the subject . the main results of the paper were previously announced in @xcite . let @xmath76 denote @xmath77 with the lorentz scalar product @xmath78 where @xmath79 , and with the space and time orientations defined , respectively , by the volume form @xmath80 and the positive light cone @xmath81 the _ mbius space _ @xmath82 is the projectivization of @xmath83 , endowed with the oriented conformal structure induced by the scalar product and the space and time orientations . if @xmath84 is the standard basis of @xmath76 , the map @xmath85\in \mathcal{m}_3.\ ] ] is an orientation - preserving conformal diffeomorphism of @xmath62 onto the mbius space minus the point @xmath86 $ ] . the inverse of @xmath87 is the conformal projection @xmath88\in \mathcal{m}_3\setminus\{p_{\infty}\}\mapsto \frac{1}{v^4}(v^1,v^2,v^3)\in \r^3.\ ] ] the _ mbius group _ @xmath22 consists of all pseudo - orthogonal transformations preserving the volume form . it is a 10-dimensional lie group with two connected components . the first component is the subgroup @xmath61 consisting of all @xmath89 preserving the positive light cone and the second one consists of all @xmath89 switching the positive light cone with the negative one . the group @xmath22 acts effectively and transitively on the left of @xmath82 preserving the conformal structure . the classical liouville theorem @xcite asserts that every conformal automorphism of @xmath82 is induced by a unique element of @xmath22 . consequently , the mbius group can be viewed as the pseudo - group of all conformal transformations of euclidean 3-space . the orientation - preserving conformal transformations are induced by the elements of @xmath61 , while the conformal transformations induced the elements of @xmath90 are orientation - reversing . for each @xmath91 , we denote by @xmath92 its column vectors . then , @xmath93 is a _ positive light cone _ basis of @xmath76 , that is a positive - oriented basis such that @xmath94 conversely , if @xmath93 is a positive light - cone basis , then the matrix @xmath72 with column vectors @xmath92 i s an element of @xmath61 . the lie algebra of @xmath22 consists of all skew - adjoint matrices of the scalar product ( [ 1.1.1 ] ) , that is @xmath95 the maximal compact abelian subgroups of @xmath22 are conjugate to the 2-dimensional torus @xmath96 where @xmath97 note that @xmath98 is the composition of the euclidean rotation of angle @xmath99 around the @xmath58-axis with the _ toroidal rotation _ of angle @xmath100 around the _ clifford circle _ @xmath101 . the @xmath58-axis and the clifford circle are the _ rotational axes _ of @xmath102 . the rotational axes of any other maximal torus @xmath103 are the images under @xmath72 of the axes of @xmath102 . let @xmath104 be a smooth curve parametrized by arclength @xmath3 , @xmath105 an open interval . points where the infinitesimal conformal arclength @xmath9 ( cf . ) vanishes are called _ vertices _ of @xmath5 ( cf . generic curves , i.e. , without vertices , can be parametrized by the _ conformal arclength _ parameter @xmath4 , defined ( up to a constant ) by @xmath106 . if such a conformal parametrization is defined for every @xmath107 , the curve is said _ complete_. a _ frame field _ along @xmath5 is a smooth map @xmath108 , such that @xmath109 . we have the following . [ p1.2.1 ] for any oriented generic curve @xmath110 , there is a unique frame field @xmath111 along @xmath5 , the _ vessiot frame _ , such that @xmath112 where @xmath113 , @xmath114 are smooth functions , called the _ conformal curvatures_. we call @xmath115 the _ canonical null lift _ of @xmath5 . if @xmath5 is biregular and @xmath116 , @xmath117 , @xmath118 is its frenet frame , with frenet serret equations @xmath119 , @xmath120 , @xmath121 , we have @xmath122 and then @xmath123{\dot{\kappa}^2+\kappa^2\tau^2}\ , ds$ ] ( cf . the conformal curvatures take the form @xmath124 where @xmath125 ( cf . @xcite ) . for a geometric description of vertices and the conformal arclength via the osculating circles of @xmath5 we refer to @xcite . the construction of @xmath126 is explicit and only involves derivatives and simplification of @xmath127 by linear relations on entries . if @xmath128 is a periodic parametrization of a closed curve , @xmath126 is a periodic @xmath61-valued map . the value of @xmath126 at @xmath129 depends on the fourth order jet of @xmath5 at @xmath130 . if @xmath131 , any solution of the linear system is the vessiot frame of the conformal parametrization @xmath132 $ ] of a generic curve with curvatures @xmath113 and @xmath114 . if @xmath72 , @xmath133 are two solutions of , with initial conditions in @xmath61 , there is a unique @xmath134 , such that @xmath135 . this shows that the conformal curvatures determine the curve , up to an orientation - preserving conformal transformation . a curve @xmath136 is called _ chiral _ if its symmetry group @xmath137 is contained in @xmath61 . a generic curve @xmath138 parametrized by conformal arclength is said _ quasi - periodic _ if its conformal curvatures are non constant , periodic , with a common minimal period @xmath139 ; @xmath140 is called the _ conformal wavelength _ of @xmath5 . the _ monodromy _ of a quasi - periodic curve is the element @xmath141 @xmath142 @xmath143 . by construction , @xmath141 satisfies @xmath144 therefore , @xmath141 generates a subgroup @xmath145 , the _ monodromy group _ of @xmath5 . if @xmath5 is closed , the integral of @xmath9 along @xmath5 is @xmath146 , where @xmath49 is the cardinality of @xmath147 . if @xmath21 is an orientation - reversing conformal transformation , the conformal curvatures of @xmath148 are @xmath149 and @xmath114 , respectively . thus , generic curves whose first curvature is nowhere vanishing are chiral . in this case , we assume that @xmath5 has _ positive chirality _ ( i.e. , @xmath150 ) . if @xmath5 is a real - analytic curve with positive chirality , then @xmath151 and @xmath5 is closed if and only if @xmath141 has finite order . note that the symmetry index of a real - analytic closed curve with positive chirality coincides with the order of the monodromy . the critical curves of the functional @xmath152 are characterized by the euler lagrange equations @xcite @xmath153 where @xmath154 is a constant of integration and @xmath155 is the second order derivative of @xmath113 with respect to the conformal arclength @xmath4 . [ e:1 ] let @xmath5 be a critical curve with constant conformal curvatures . then , either @xmath156 and @xmath157 , or @xmath158 and @xmath159 . in the second case , we may assume @xmath150 . the class of curves with constant conformal curvatures was studied in @xcite : they are equivalent to the rhumb lines of either a torus of revolution , or a round cone , or else a circular cylinder . since we deal with curves without vertices , the meridians and the parallels must be excluded from the discussion . the rhumb lines of a round cone , which possibly can degenerate into the punctured plane , are helices over logarithmic spirals , while the rhumb lines of a circular cylinder are circular helices . all of them are not closed . from the viewpoint of the conformal geometry , any rotationally invariant torus is equivalent to a torus @xmath160 generated by rotating around the @xmath58-axis the circle in the @xmath161-plane with radius @xmath162 and center @xmath163 , for some @xmath164 . the latter are the regular orbits of the action of @xmath102 on the euclidean space . if @xmath165 are parametric equations of @xmath160 , its rhumb lines are @xmath166 where @xmath167 and @xmath168 . the conformal curvatures of @xmath169 are @xmath170{8m^2n^2(m^2-n^2)^2r^2(2-r^2)^2}},\\ k_2=\frac{(8n^4r^2+m^4(r^2 - 2)^2)\mid mn(m^2-n^2)\mid}{4\sqrt{2}rm^2n^2(m^2-n^2)(r^2 - 2)}. \end{cases}\ ] ] they verify the inequality @xmath171 . vice versa , every curve with constant conformal curvatures that satisfies the above inequality is equivalent to a rhumb line of @xmath160 , for some @xmath164 . possibly , @xmath172 can be computed in terms of the constants @xmath113 and @xmath114 . from this it follows that , up to conformal transformations , the only closed critical curves with constant conformal curvatures are of the form @xmath173 , where @xmath32 is a positive rational number different from @xmath174 and @xmath175 the trajectory of @xmath173 is a torus knot of type @xmath176 , where @xmath177 are coprime integers such that @xmath178 , see figure [ fig1 ] . and @xmath179 , respectively.,title="fig:",width=226,height=226 ] and @xmath179 , respectively.,title="fig:",width=226,height=226 ] from now on we consider critical curves with non - constant conformal curvatures , parametrized by the conformal arclength parameter . then , implies @xmath180 where @xmath181 is another constant of integration . this equation has non - constant periodic solutions if and only if the polynomial @xmath182 has two distinct real roots , denoted by @xmath183 and ordered in such a way that @xmath40 with @xmath39 and @xmath184 . then can be rewritten in the form @xmath185 the general solution of ( [ 2.1.3 ] ) is given by @xmath186 where @xmath187 , @xmath188 is any real constant , and @xmath189 is the jacobi elliptic function @xmath190 by possibly reversing the orientation along the curve , acting with an orientation - reversing conformal transformation and shifting the independent variable , we can assume that @xmath191 and @xmath192 . we have proved the following . [ prop1 ] the mbius classes of quasi - periodic critical curves are in 1 - 1 correspondence with the points @xmath193 of the planar region @xmath194 we call @xmath37 and @xmath38 the _ parameters _ of a quasi - periodic critical curve . the mbius class of a critical curve with parameters @xmath37 and @xmath38 is represented by any conformal parametrization @xmath136 with curvatures @xmath195 note that the minimal period of @xmath189 is given by @xmath196 where @xmath197 is the complete elliptic integral of the first kind . let @xmath198 denote the open subset of @xmath199 defined by @xmath200 for every @xmath201 , let @xmath202 and define @xmath203 these integrals can be evaluated in term of the complete integral of the third kind @xmath204 as a result , we have @xmath205 we call @xmath206 the _ period map _ of the conformal arclength functional . by construction , @xmath207 is non constant and real analytic . we are now in a position to state the following . [ thm1 ] a quasi - periodic critical curve with parameters @xmath37 and @xmath38 is a conformal string if and only if @xmath201 and @xmath208 , @xmath209 . for every @xmath210 , let @xmath211 be the conformal parametrization of a quasi - periodic critical curve with parameters @xmath210 , such that @xmath212 , where @xmath72 is the vessiot frame along @xmath5 . the claim is that @xmath5 is periodic if and only if @xmath201 and @xmath213 , @xmath214 . for brevity , let @xmath215 denote the first conformal curvature @xmath113 , @xmath216 and @xmath217 denote the constants @xmath218 , @xmath219 , respectively , and @xmath140 denote the minimal period @xmath220 . [ l : momentum ] the canonical null lift @xmath221 of @xmath5 satisfies the linear system @xmath222 with the initial condition @xmath223 , where @xmath224 using lemma [ l : momentum ] , an explicit integration of the critical curves was carried out in @xcite ( see also @xcite ) . the proof of theorem [ thm1 ] is a direct consequence of this integration . according to the analysis carried out in @xcite , we are led to consider three cases , depending on whether @xmath225 , @xmath226 , or @xmath42 . it follows from @xcite that the conformal parametrizations of a quasi - periodic critical curve with parameters @xmath37 , @xmath38 can only be periodic if @xmath42 , in which case the following holds . [ rl ] a quasi - periodic critical curve with @xmath42 is closed if and only if @xmath208 and @xmath227 are rational numbers . assume that @xmath39 , @xmath40 , and @xmath42 . then , the eigenvalues of @xmath228 are @xmath229 , @xmath230 , and @xmath231 . choose @xmath232 , such that @xmath233 and consider the map @xmath234 . from ( [ 2.2.4 ] ) it follows that the components @xmath235 of @xmath236 satisfy @xmath237 using @xmath238 , the above equations take the form @xmath239 thus @xmath240 , where @xmath241 and @xmath242 are the complex - valued functions given by @xmath243 since @xmath215 is a non - constant periodic function with minimal period @xmath140 , the functions @xmath242 are periodic if and only if @xmath244 which proves the lemma . this proves theorem [ thm1 ] . as a corollary , we have the following . [ prop2 ] the mbius classes of conformal strings are in 1 - 1 correspondence with the elements of the countable set @xmath245 we will obtain theorem [ thm : a ] from proposition [ prop2 ] and the next result about the period map @xmath207 . [ thm2 ] the period map @xmath246 is a real - analytic diffeomorphism onto the domain @xmath247 the remaining part of the section is taken up by a detailed study of the analytic properties of the transformation @xmath207 , from which theorem [ thm2 ] follows at once . note that @xmath248 and @xmath249 extend analytically to the domain @xmath250 we will use the same notation for @xmath248 , @xmath249 and their extensions on @xmath251 . let @xmath252 be the elliptic integral of the second kind . then , @xmath253 and the ratio @xmath254 is a strictly decreasing function from @xmath255 onto @xmath256 $ ] . the power series of @xmath102 and @xmath257 ( cf . @xcite , p. 73 ) are given by @xmath258 we also recall the expressions of the derivatives of the complete elliptic integral of the third kind @xmath259 ( cf . ) , @xmath260 using ( [ 2.2.3 ] ) and ( [ 4.3 ] ) , we have @xmath261 where the coefficients @xmath262 , @xmath263 , and @xmath264 are given by @xmath265 and by @xmath266 in the formulae above , @xmath267 stands for @xmath268 . these formulae have been derived with the help of the software _ mathematica_. we now prove the following . [ prop4.i ] the partial derivatives of @xmath248 and @xmath249 are strictly positive on @xmath269 we begin by proving the following . [ lemma4i ] @xmath270 , for every @xmath271 . take @xmath272 and @xmath273 $ ] . set @xmath274 , with @xmath275 , and put @xmath276 . from ( [ 4.4])@xmath277 , we have @xmath278 where @xmath279 first , we prove that @xmath280 , for every @xmath281 . if @xmath281 , then @xmath282 . so , if we set @xmath283 , it suffices to show that @xmath284 . the derivative of @xmath285 with respect to @xmath37 is @xmath286 then , @xmath287 is a strictly decreasing function such that @xmath288 . this implies @xmath289 computing the derivative of the function @xmath290 , we find @xmath291 then , @xmath292 is strictly increasing and @xmath293 , for every @xmath281 . this implies @xmath280 , for every @xmath281 . 0.1 cm next , we prove that @xmath294 . using the power series expansions of @xmath295 and @xmath296 , we find @xmath297 from this , we have @xmath298 by similar arguments as in the proof of lemma [ lemma4i ] , one can prove that , for every @xmath271 , @xmath299 , @xmath300 , and @xmath301 , which completes the proof of proposition [ prop4.i ] . we will now prove the following . [ prop4.ii ] the jacobian of @xmath207 is strictly positive on @xmath198 . in particular , the image @xmath302 is a connected open set and @xmath303 is a local diffeomorphism . let @xmath304 be the matrix of the differential of @xmath207 . using ( [ 4.4 ] ) , the jacobian @xmath305 of @xmath207 at @xmath193 is given by @xmath306 let @xmath307 denote the numerator of the right hand side . if we let @xmath274 , where @xmath281 and @xmath308 , then @xmath309 which implies the required result . we adopt the following conventions : * @xmath310 is considered as a function defined on @xmath198 ; * @xmath311 stands for the extension of @xmath207 to @xmath312 ; * @xmath313 denotes the restriction of @xmath314 to the closure @xmath315 of @xmath198 . next , we prove the following . [ prop4.iii ] the mapping @xmath207 is a real - analytic local diffeomorphism onto @xmath316 . by proposition [ prop4.ii ] , it suffices to prove that @xmath316 is the image of @xmath207 . the boundary of @xmath198 consists of the simple arcs @xmath317 which intersect at the point @xmath318 , while the boundary of @xmath316 is made of the three simple arcs @xmath319\},\\ & \partial_0 \widetilde{\omega } = \{(-1/2,b ) : b\in [ 0,1/2]\},\\ & \partial_{+}\widetilde{\omega } = \{(a,\sqrt{a^2 - 1/2 } ) : a\in [ -1/\sqrt{2},-1/2]\ } \end{aligned}\ ] ] with vertices @xmath320 , @xmath321 and @xmath322 , respectively . the restriction @xmath323 of @xmath313 to @xmath324 is given by @xmath325 it is then a diffeomorphism of @xmath326 onto @xmath327 . similarly , the restriction @xmath328 of @xmath313 to @xmath329 is the diffeomorphism onto @xmath330 given by @xmath331 then @xmath313 maps the boundary of @xmath198 to the boundary of @xmath316 . next , we show that the image of @xmath207 is contained in @xmath316 . for , fix @xmath272 and consider the curve defined by @xmath332 , for every @xmath333 . from proposition [ prop4.i ] , we know that the components @xmath334 and @xmath335 of @xmath336 are increasing functions . this implies @xmath337 and @xmath338 combining these two inequalities , we find @xmath339 this shows that the image of @xmath207 is contained in @xmath316 . to prove equality , we need the following technical lemma . [ 4.v ] let @xmath340 be a sequence such that @xmath341 . then , @xmath342 . the trajectory of @xmath336 is the graph of an increasing function @xmath343 given the sequence @xmath344 , we write @xmath345 , where @xmath346 is an element of the open interval @xmath347 . from the limits @xmath348 we have @xmath349 by proposition [ prop4.ii ] , @xmath350 is a real - analytic local diffeomorphism onto its image , and hence an open map . consequently , @xmath351 is a closed subset of @xmath316 . the proof is complete if we show that @xmath352 is also open . suppose it is not open . then , there is a point @xmath353 , such that any open disk @xmath354 centered at @xmath355 with radius @xmath356 , @xmath357 , intersects @xmath302 . for every @xmath357 , choose @xmath358 and @xmath359 , such that @xmath360 . two possibilities may occur : either @xmath361 is bounded , or else @xmath361 is unbounded . in the first case , we may assume that @xmath361 converges to a limit point @xmath362 . by construction , @xmath362 belongs to @xmath315 . if @xmath362 is an element of @xmath198 , then @xmath363 , which contradicts the fact that @xmath364 . so , @xmath362 is an element of the boundary @xmath365 . this implies that @xmath366 . on the other hand , @xmath313 maps the boundary of @xmath198 onto the boundary of @xmath316 . consequently , @xmath355 would be an element of @xmath367 , which is absurd , since @xmath368 . if @xmath361 is unbounded , by lemma [ 4.v ] , the sequence made up with the abscissae of the points @xmath369 converges to @xmath370 . so , the first coordinate of @xmath355 would be equal to @xmath370 and hence @xmath371 , which is a contradiction . this concludes the proof of proposition [ prop4.iii ] . by proposition [ prop4.i ] , we know that the first order partial derivatives of @xmath248 and @xmath249 are strictly positive on @xmath372 . then , there is an open neighborhood @xmath373 of @xmath372 such that the first order partial derivatives of @xmath248 and @xmath249 are positive on @xmath374 . on this set , consider the nowhere vanishing vector fields ( see figure [ fig2 ] ) @xmath375 and @xmath376 , with some of their integral curves.,title="fig:",width=226,height=226 ] and @xmath376 , with some of their integral curves.,title="fig:",width=226,height=226 ] by construction , the trajectories of the integral curves of @xmath377 and @xmath378 are graphs of strictly decreasing functions , and hence they intersect @xmath379 in at most one point . in addition , for every @xmath380 , we have * the connected components of the level curve @xmath381 are contained in the intersection of a trajectory of @xmath377 with @xmath198 . * the connected components of the level curve @xmath382 are contained in the intersection of a trajectory of @xmath378 with @xmath198 . we now prove the following . [ 4.vi ] the level curve @xmath383 is connected , for every @xmath384 . consider the arcs @xmath385 , @xmath386 , @xmath387 and @xmath388 . from the proof of proposition [ prop4.iii ] , we know that the maps @xmath389 and @xmath390 are real - analytic diffeomorphisms . for every @xmath384 , let @xmath391 denote the intersections of @xmath392 with the line @xmath393 and let @xmath394 and @xmath395 be the points of @xmath396 , such that @xmath397 . note that @xmath398 . 0.1 cm @xmath399 first , we prove that @xmath400 is compact . by continuity , @xmath400 is closed . suppose that @xmath400 is unbounded . then , there is a sequence @xmath401 , such that @xmath402 . by lemma [ 4.v ] , @xmath403 converges to @xmath370 . on the other hand , the points @xmath404 belong to @xmath383 , and hence @xmath405 , which is the limit of the sequence @xmath403 , must be equal to @xmath370 . but this is absurd , since @xmath406 and the abscissae of the points in @xmath316 are different from @xmath370 . this proves the compactness of @xmath400 . 0.1 cm @xmath399 next , we prove that the points @xmath407 belong to the boundary of any connected component of the level curve @xmath383 . let @xmath408 be a connected component of @xmath383 . we know that @xmath408 is the graph of a strictly decreasing function and that such a function is bounded , by the compactness of @xmath400 . since @xmath408 is an embedded curve , its closure consists of two distinct points . on the other hand , @xmath409 . using the fact that @xmath410 is an embedded curve , we have that @xmath411 and @xmath412 are the two boundary points of @xmath408 . 0.1 cm @xmath399 to complete the proof , we show that two connected components of the level curves intersect each other . the vector field @xmath377 does not vanishes at the point @xmath411 and @xmath411 belongs to the boundary of @xmath408 . therefore , @xmath408 is contained in the trajectory of @xmath377 passing through @xmath411 . choose local coordinates @xmath413 defined on a cubical open neighborhood @xmath21 of the point @xmath411 , such that @xmath414 and @xmath415 . in these coordinates , the intersection @xmath416 consists of the points @xmath417 such that @xmath418 and @xmath419 , for some @xmath420 . so , if @xmath408 and @xmath421 are two connected components of @xmath383 , there exist @xmath422 , such that @xmath423 and @xmath424 . this implies that @xmath425 is not empty . a similar result also holds for the other family of level curves . the level curves @xmath426 are connected , for every @xmath384 . let @xmath408 be any connected component of @xmath426 . fix @xmath427 and take a strictly decreasing function @xmath428 whose graph coincides with @xmath408 . such a function is defined on an open interval @xmath105 of the form @xmath429 , where @xmath430 and @xmath431 . let @xmath432 be a decreasing sequence , such that @xmath433 . by construction , @xmath434 is an increasing sequence and @xmath435 , for every @xmath49 . in particular , @xmath436 , which implies that @xmath434 is bounded . by possibly taking a subsequence , we may assume that @xmath434 converges to a limit point @xmath437 . since @xmath438 converges to @xmath439 , the point @xmath439 belongs to @xmath440 . 0.1 cm @xmath399 we prove that @xmath439 is an element of @xmath441 . by contradiction , suppose that @xmath442 . then , @xmath408 is contained in the trajectory , @xmath421 , of the vector field @xmath443 passing through @xmath439 . by definition , @xmath444 and @xmath445 . therefore , @xmath421 is a connected subset of @xmath426 which contains properly @xmath408 . this is absurd since @xmath408 is a connected component of @xmath426 . 0.1 cm by construction , since @xmath446 and @xmath447 is an increasing sequence , @xmath448 . this shows that @xmath439 does not belong to @xmath449 . therefore , @xmath439 is an element of @xmath324 and hence @xmath450 and @xmath451 . the point @xmath452 is independent of the choice of the connected component . otherwise , the restriction of @xmath311 to @xmath324 could not be injective . we are now in the same situation as in the last part of the proof of the previous lemma , i.e. , the point @xmath452 belongs to the boundary of any connected component of the level curve @xmath426 . by arguing as above , we deduce the result . we are now ready to prove the last ingredient for the proof of theorem [ thm2 ] . [ prop4.iv ] the mapping @xmath207 is injective . it suffices to show that @xmath453 , for every @xmath454 . by contradiction , suppose the existence of @xmath454 , such that @xmath455 . let @xmath193 and @xmath456 be two distinct elements of @xmath198 , such that @xmath457 @xmath458 @xmath459 @xmath458 @xmath460 . from the above discussions , we know that the level curves @xmath461 and @xmath462 are connected and graphs of two strictly decreasing functions , denoted by @xmath463 and @xmath464 , respectively . the domain of definition is an open interval @xmath465 , containing @xmath37 and @xmath466 . by construction , @xmath467 , @xmath468 , with @xmath469 . on the other hand , @xmath461 and @xmath462 are contained in the trajectories of the vector fields @xmath377 and @xmath378 , respectively . from this , we have @xmath470 proposition [ prop4.ii ] implies @xmath471 then , the function @xmath472 satisfies @xmath473 , @xmath474 and @xmath475 . this implies the existence of @xmath476 , different from @xmath37 and @xmath466 , such that @xmath477 and @xmath478 . consequently , the jacobian of @xmath207 is non positive at @xmath479 @xmath458 @xmath480 , contrary to proposition [ prop4.ii ] . we have just proved that the mbius classes of conformal strings are in 1 - 1 correspondence with the _ moduli _ , i.e. , the elements of the countable set @xmath481 then , for every @xmath482 , there is a unique @xmath201 such that @xmath483 and @xmath484 . for every @xmath485 , let @xmath486 and @xmath487 where @xmath183 are the parameters of @xmath32 and @xmath215 , @xmath216 , @xmath217 stand for @xmath488 , @xmath218 and @xmath219 , respectively . note that @xmath489 and @xmath193 are related by @xmath490 where @xmath491 is the minimal period of @xmath215 . the _ symmetrical configuration _ of the conformal strings with modulus @xmath492 is the parametrized curve @xmath493 , defined by @xmath494 theorem [ thm : b ] is a consequence of the following . [ 5.i ] let @xmath492 be any element of @xmath495 and @xmath496 be the corresponding parameters . any conformal string with parameters @xmath193 is conformally equivalent to the symmetrical configuration @xmath54 . consider the unique conformal parametrization @xmath136 of a conformal string with parameters @xmath37 , @xmath38 whose vessiot frame @xmath72 satisfies the initial condition @xmath212 . it suffices to prove that @xmath5 is conformally equivalent to @xmath497 . by lemma [ rl ] , the canonical lift @xmath498 of @xmath5 takes the form @xmath499 , where @xmath500 and @xmath501 is the @xmath502-valued map defined by @xmath503 on the other hand , the curve @xmath504 , defined by @xmath505 is a null lift of @xmath54 . from and , it follows that @xmath506 , for some @xmath507 . consequently , @xmath508 , for a suitable @xmath509 . this yields @xmath510 , where @xmath511 is the vessiot frame along @xmath54 . thus @xmath512 , which implies that @xmath5 and @xmath54 are equivalent to each other . the map @xmath207 can be inverted by numerical methods . once we know the parameters @xmath37 and @xmath38 which correspond to the modulus @xmath32 , we can use ( [ 5.4 ] ) to find the explicit parametrization of the symmetrical configuration @xmath54 . for every @xmath482 , let @xmath45 and @xmath46 , with @xmath513 . let @xmath49 denote the least common multiple of @xmath50 and @xmath51 . then , @xmath514 and @xmath515 , where @xmath516 and @xmath517 are two positive coprime integers . we are now ready to prove theorem [ thm : c ] . the curve @xmath5 is a real - analytic closed curve with positive chirality . therefore , its symmetry group is generated by the monodromy . by construction , the canonical null lift @xmath498 is as in ( [ 5.5 ] ) and the first conformal curvature is a strictly positive periodic function , with minimal period @xmath140 . using , we find @xmath518 , where @xmath519 is as in . then , @xmath520 is the monodromy of @xmath5 . this implies that the symmetry group of @xmath5 is generated by @xmath520 . let @xmath521 $ ] denote the @xmath58-axis with the downward orientation induced by the parametrization @xmath522 , and let @xmath523 $ ] be the clifford circle equipped with the orientation induced by the rational parametrization @xmath524 now we compute the gauss linking integrals @xmath525)$ ] and @xmath526)$ ] . if @xmath527 , the gauss linking integral @xmath525)$ ] is given by @xmath528)&=\frac{1}{4\pi } \int_{[\gamma]}\int_{z}\frac{\gamma-\alpha}{\\|\gamma-\alpha\\|^3}\cdot d\gamma\times d\alpha\\ & = -\frac{1}{4\pi } \int_{0}^{n\omega}\left(\int_{-\infty}^{+\infty}\left(\frac{x(t)y'(t)-x'(t)y(t)}{(x(t)^2 + y(t)^2+(z(t)-s)^2)^{3/2}}\right)ds\right)dt\\ & = -\frac{1}{2\pi } \int_0^{n\omega}\frac{x(t)y'(t)-x'(t)y(t)}{x(t)^2+y(t)^2}dt\\ & = -\frac{n}{2\pi}\int_0^{\omega}\theta_2'(t)dt = n\phi_2= n q_2 = h_2m_2 . \end{split}\ ] ] to compute the linking integral of @xmath5 with the clifford circle we consider the orientation - preserving conformal involution @xmath529 this map takes @xmath530 to the point at infinity and exchanges the roles of the two axes of symmetry , i.e. , @xmath531 . a direct computation shows that the parametric equations of @xmath532 are @xmath533 where @xmath534 . we then have @xmath535)&=-\mathrm{lk}(\gamma_,[z])= -\frac{1}{4\pi } \int_{[\gamma_]}\int_{z}\frac{\gamma_-\alpha}{\\|\gamma_-\alpha\\|^3}\cdot d\gamma_\times d\alpha\\ & = \frac{1}{4\pi } \int_{0}^{n\omega}\left(\int_{-\infty}^{+\infty}\left(\frac{x_(t)y _ * ' ( t)-x _ * ' ( t)y_(t)}{(x_(t)^2+y_(t)^2 + ( z_(t)-s)^2)^{3/2}}\right)ds\right)dt\\ & = \frac{1}{2\pi } \int_0^{n\omega}\frac{x_(t)y _ ' ( t)- x _ * ' ( t)y_(t)}{x_(t)^2+y_*(t)^2}dt\\ & = \frac{n}{2\pi}\int_0^{\omega}\theta_1'(t)dt = -n\phi_1= nq_1 = h_1m_1 , \end{split}\ ] ] which yields the required result . , let @xmath5 be the symmetrical configuration with modulus @xmath492 . then , @xmath537 is the euclidean rotation of angle @xmath538 around the @xmath58-axis . the subgroup of order @xmath516 generated by @xmath539 is the _ euclidean symmetry group _ of the symmetrical configuration . similarly , @xmath540 is the toroidal rotation of angle @xmath541 around the clifford circle . the cyclic subgroup of order @xmath517 generated by @xmath542 is the _ clifford symmetry group _ of the symmetrical configuration . denoting by @xmath140 the conformal wavelength of @xmath5 , the arc @xmath18_0=\gamma([0,\omega))\subset [ \gamma]$ ] is the _ fundamental domain _ of the string and the trajectory @xmath18 $ ] is obtained from @xmath18_{0}$ ] via @xmath26 , i.e. , @xmath543=\bigcup_{a\in g_{\gamma } } a\cdot [ \gamma]_0 $ ] . , @xmath544 , and @xmath545 , respectively.,title="fig:",width=226,height=226 ] , @xmath544 , and @xmath545 , respectively.,title="fig:",width=226,height=226 ] , @xmath544 , and @xmath545 , respectively.,title="fig:",width=226,height=226 ] the symmetrical configurations are labeled by three quantum numbers : the order of the symmetry group and the linking numbers with the clifford circle and the @xmath58-axis . we use the notation @xmath546 for the symmetrical configuration with a symmetry group of order @xmath49 and linking numbers @xmath526)=\ell_1 $ ] and @xmath525)=\ell_2 $ ] , respectively . let @xmath59 be the cardinality of the set of the symmetrical configurations with index of symmetry @xmath49 . there are no symmetrical configurations with @xmath547 . figure [ fig3 ] reproduces the graph of the function @xmath548 , for @xmath549 . this suggest that @xmath550 has an asymptotic quadratic growth . in this final example , we consider the symmetrical configurations with @xmath552 . this set consists of five elements : @xmath553 , @xmath544 , @xmath545 , @xmath554 , and @xmath551 . the moduli of these strings are @xmath555 , @xmath556 , @xmath557 , @xmath558 , and @xmath559 , respectively . the euclidean and the clifford symmetry subgroups of the first two strings are trivial . the euclidean symmetry group of the third string is trivial , while the clifford symmetry group has order 3 ( see figure [ fig4 ] ) . the fourth and fifth strings have an euclidean symmetry group of order 3 and have no non - trivial clifford symmetries ( see figure [ fig5 ] ) . the invariants @xmath37 , @xmath38 and the conformal wavelength @xmath140 can be tabulated against the quantum numbers @xmath49 , @xmath560 and @xmath561 . for instance , the values of the natural parameters @xmath37 , @xmath38 and those of the conformal wavelength @xmath140 of the stings with @xmath552 are @xmath562 the shape of the strings becomes more complicated when @xmath49 , @xmath560 and @xmath561 increase , see for instance figure [ fig6 ] , where the symmetrical configuration with @xmath563 , @xmath564 and @xmath565 is reproduced . this string has no non - trivial euclidean symmetries and the subgroup of its clifford symmetries has order 3 .	by a _ conformal string _ in euclidean space is meant a closed critical curve with non - constant conformal curvatures of the conformal arclength functional . we prove that ( 1 ) the set of conformal classes of conformal strings is in 1 - 1 correspondence with the rational points of the complex domain @xmath0 and ( 2 ) any conformal class has a model conformal string , called _ symmetrical configuration _ , which is determined by three phenomenological invariants : the order of its symmetry group and its linking numbers with the two conformal circles representing the rotational axes of the symmetry group . this amounts to the quantization of closed trajectories of the contact dynamical system associated to the conformal arclength functional via griffiths formalism of the calculus of variations . = 1
let @xmath1 be a real algebraic cubic surface in @xmath0 and let @xmath2 be a real algebraic quadric surface in @xmath0 . this just means that for some homogeneous polynomials @xmath3 , @xmath4 , and @xmath5 , @xmath6 , surfaces @xmath1 and @xmath2 are given by the following equations @xmath7\in{\bbb r}p^3\ \|\ f(x_0,x_1,x_2,x_3)=0\}\\ { \bbb r}q=\{[x_0:x_1:x_2:x_3]\in{\bbb r}p^3\ \|\ g(x_0,x_1,x_2,x_3)=0\}.\end{gathered}\ ] ] let @xmath8 and @xmath9 be the complexifications of surfaces @xmath1 and @xmath2 @xmath10\in{\bbb c}p^3\ \|\ f(x_0,x_1,x_2,x_3)=0\}\\ { \bbb c}q=\{[x_0:x_1:x_2:x_3]\in{\bbb c}p^3\ \|\ g(x_0,x_1,x_2,x_3)=0\}.\end{gathered}\ ] ] we suppose that @xmath8 and @xmath9 are nonsingular and intersect transversely along a curve @xmath11 . the intersection @xmath12 is a smooth curve in @xmath13 , its real part @xmath14 is @xmath15 . consider now the diffeomorphism type of @xmath8 , @xmath9 , @xmath11 . it is well - known that @xmath9 is diffeomorphic to @xmath16 , @xmath8 is diffeomorphic to @xmath17 and we can construct @xmath8 from @xmath18 by blowing up 6 points not sitting on the same conic . the curve @xmath11 is the proper transform of a plane curve of degree 6 with 6 ordinary double points in the blowup points so the genus of @xmath11 is 4 . consider now the diffeomorphism type of @xmath1 , @xmath2 , @xmath14 . it is well - known that a quadric @xmath2 is a hyperboloid ( i.e. diffeomorphic to @xmath19 ) , an ellipsoid ( i.e. diffeomorphic to @xmath20 ) or empty . it is also well - known that a cubic @xmath8 is diffeomorphic to @xmath21 , where @xmath22 or @xmath23 . since @xmath14 is smooth , every component of @xmath14 is diffeomorphic to @xmath24 . note that the embedding of this @xmath24 into @xmath1 is two - sided , the sides of @xmath14 are defined by the sign of polynomial @xmath5 ( since @xmath25 is even ) . since the genus of @xmath11 is 4 , the harnack inequality implies that the number of components of @xmath14 is at most 5 ( @xmath26 is a connected surface and , therefore , the number of components of its boundary is at most @xmath27 ) . the topological classification of pairs @xmath28 was done by gudkov @xcite in the case where @xmath2 is a hyperboloid and by gudkov and shustin @xcite in the case where @xmath2 is an ellipsoid ( see @xcite for the classification of flexible curves ) . this paper gives the topological classification of pairs @xmath29 . two new phenomena appear in this classification the arrangement of @xmath14 with respect to the components of @xmath1 in the case when @xmath1 is not connected and the arrangement of the components of @xmath14 with respect to the handles of @xmath1 in the case when @xmath1 is diffeomorphic to a projective plane with handles . these possibilities for arrangement turn out to be controlled by the degree of the curve . the topological type of a pair @xmath29 is also called the type of _ topological arrangement _ of @xmath14 in @xmath1 or the _ real scheme _ of @xmath14 in @xmath1 . a part of the work was done during my stay at the institute for advanced study ( supported by the nsf grant dms 9304580 ) and at the max - planck - institut . there are two diffeomorphism types of cubic surfaces of positive euler characteristic in @xmath0 : @xmath30 we consider them separately since if @xmath1 is one of these surfaces than any component of @xmath14 bounds a disk in @xmath1 and even the notations for the topological type of @xmath29 are simpler in this case . recall the system of codes for topological arrangements of a collection of two - sided circles ( or _ ovals _ ) in @xmath31 introduced by viro @xcite . every oval separates @xmath31 into a disk ( caller the _ interior _ of the oval ) and a mbius band . two ovals are called _ disjoint _ if their interiors are disjont . the system of codes is inductive , @xmath32 disjoint ovals are encoded by @xmath33 and if we add a new oval containing a collection @xmath34 in its interior then the code of the new collection is @xmath35 . to make the same system suitable for encoding of topological arrangements of embedded curves in @xmath20 it suffices to fix a point in @xmath20 in the complement of the curve then the interior of an oval is defined as the component of the complement of the oval not containing the fixed point . such a code determines uniquely a topological type of curve in @xmath20 , but because of ambiguity in fixing a point there is more than one code corresponding to the same topological type . we choose codes minimizing the number of ovals in the interior of other ovals . to encode types of topological arrangement of @xmath14 into a cubic surface diffeomorphic to @xmath23 we combine systems for @xmath31 and @xmath20 . parameters @xmath36 and @xmath37 in the theorems are nonnegative integer numbers . [ hyp ] topological arrangement of a nonsingular algebraic curve of degree 6 on a nonsingular cubic surface diffeomorphic to @xmath23 is of one of the following types : a : : @xmath38 b : : @xmath39 c : : @xmath40 d : : @xmath41 e : : @xmath42 f : : @xmath43 each of the 31 types listed above is realizable by a nonsingular algebraic curve of degree 6 . [ rp2 ] topological arrangement of a nonsingular algebraic curve of degree 6 on a nonsingular cubic surface diffeomorphic to @xmath31 is of one of the following types : a : : @xmath44 b : : @xmath45 c : : @xmath46 . each of the 17 types listed above is realizable by a nonsingular algebraic curve of degree 6 . there are three diffeomorphism types of cubic surfaces of negative euler characteristic in @xmath0 : @xmath47 we define @xmath48 then @xmath49 . note that at least one of @xmath50 , @xmath51 is non - orientable since @xmath52 changing the sign of @xmath5 we may assume that @xmath53 where @xmath54 stands for the number of components ( the 0-dimensional betti number ) . then @xmath51 is always nonorientable . note that the topological type of @xmath29 is determined by the topological types of @xmath50 and @xmath51 if the number of non - disk components is not more than one for both @xmath50 and @xmath51 ( since then there is only one way to glue @xmath50 and @xmath51 along the boundary to obtain a connected @xmath1 ) . in this case we code the topology of @xmath14 on @xmath1 with @xmath55 where @xmath56 is the number of disks in @xmath50 , @xmath32 is the number of disks in @xmath51 , @xmath57 is the topological type of the only non - disk component of @xmath50 ( if any ) and @xmath58 is the only non - disk component of @xmath51 . we denote the topological type of the connected orientable surface of genus @xmath5 with @xmath59 and that of the non - orientable one with @xmath60 . notation @xmath61 stands for the result of puncturing of a closed surface @xmath57 @xmath62 times . lemma [ arnold ] implies that if @xmath50 or @xmath51 has more than one non - disk components then the topological type of @xmath29 is still determined by the topological types of @xmath50 , @xmath51 and @xmath1 . if @xmath63 , @xmath64 then the code of @xmath29 is @xmath65 if the core of the annulus is null - homologous and @xmath66 if not . [ 7rp2 ] topological arrangement of a nonsingular algebraic curve of degree 6 on a nonsingular cubic surface diffeomorphic to @xmath67 is of one of the following types : @xmath68 , + [ cols=">,>,>,>,>",options="header " , ] + to construct the arrangements marked by * in the table we start from a plane curve of degree 6 and collapse 6 of its empty ovals to points using @xcite . to construct @xmath69 and @xmath70 we use the gudkov curve @xmath71 constructed in @xcite . to get @xmath69 we collapse the 5 inner ovals and one of the outer ovals ; to construct @xmath70 we collapse the 5 outer ovals and one of the inner ovals ( see fig . [ gud ] ) . to construct @xmath72 we collapse the 6 outer ovals of @xmath73 . + * @xmath74 , @xmath75 + to construct the plane curves of degree 6 with 6 double points for these arrangements we start from the union of a two - component cubic in @xmath31 and its `` parallel '' copy which intersect in 9 disjoint point and smooth 3 out of the 9 nodes of the union ( see fig . [ arn ] ) . lemma [ arnold ] implies that unless @xmath29 is of type 6 of theorem 4 or of type 4 of theorem 5 it is of type @xmath76 in the case of theorem 4 @xmath77 and in the case of theorem 5 @xmath78 since @xmath79 . to finish the restrictions for theorem 4 and theorem 5 we have to prove that @xmath80 is not realizable ( for theorem 4 ) . note that @xmath81 ( see @xcite ) is orientable and therefore @xmath82 ( @xmath83 is negative definite ) . on the other hand @xmath84 . if @xmath85 or @xmath86 is constructed in [ thm3con ] out of polotovskii s catalog then we can construct @xmath87 out of a similar pair of a conic and a quartic which intersect in 6 real points . considering similar pairs of a conic and a quartic which intersect in 4 real points produces the arrangements for theorem 5 . a similar modification works for @xmath88 and @xmath89 . this produces all the arrangements for theorem 4 except for @xmath90 and @xmath91 . to construct them we perturb a union of a quartic and two lines as shown on fig . [ arr ] . 99 arnold , v.i . : on the distribution of the ovals of real plane curves , involutions of 4-dimensional smooth manifolds and the arithmetics of integer - valued quadratic forms . analiz i ego prilozheniya , * 5(3 ) * , 1 - 9 ( 1971 ) . gudkov , d.a . : construction of the new series of m - curves . doklady academii nauk sssr , ser . mat . , * 200(6 ) * , 1269 - 1272 ( 1971 ) . gudkov , d.a . : on the topology of the algebraic curves on a hyperboloid . uspehi mat . nauk , * 34(6 ) * , 26 - 32 ( 1979 ) . gudkov , d.a . , krakhnov , a.d . : on the periodicity of euler characteristic of real algebraic ( m-1)-manifolds . functsionalni analiz i ego prilozheniya , * 7(2 ) * , 15 - 19 ( 1973 ) . gudkov , d.a . , utkin , g.a . : nine papers on hilbert s 16th problem . ams translations , series 2 , * 112 * ( 1978 ) . gudkov , d.a . , shustin , e.i . : the classification of the non - singular curves of order 8 on an ellipsoid . metody kachestvennoi teorii differentsialnyh uravneni , gorki , 104 - 107 ( 1980 ) ( russian ) . itenberg , i.v . : plane projective real curves of degree 6 with a non - degenerate double point . dissertation . 1991 ( russian ) . kharlamov , v.m . : new congruences for the euler characteristic of real algebraic manifolds . functsionalni analiz i ego prilozheniya , * 7(2 ) * , 74 - 78 ( 1973 ) . letizia , m. : quotients by complex conjugation of nonsingular quadrics and cubics in @xmath92 defined over @xmath93 . pacific j. math . , * 110 * , 307 - 314 ( 1984 ) . mikhalkin , g.b . : congruences for real algebraic curves on an ellipsoid . zapiski nauchnyh seminarov lomi , * 193 * , 90 - 100 ( 1991 ) . mikhalkin , g.b . : the complex separation of real surfaces and extensions of rokhlin congruence . , * 118 * , 197 - 222 ( 1994 ) . polotovskii , g.m . : catalog of m - decomposing curves of order 6 . doklady academii nauk sssr , ser . mat . , * 236(3 ) * , 548 - 551 ( 1977 ) . rokhlin , v.a . : congruences modulo 16 in sixteenth hilbert problem . analiz i ego prilozheniya , * 6(4 ) * , 58 - 64 ( 1972 ) . rokhlin , v.a . : complex topological characteristics of real algebraic curves . uspehi mat . nauk , * 33(5 ) * , 77 - 89 ( 1978 ) . viro , o.ya . : the progress in topology of real algebraic varieties over last six years . uspehi mat . nauk , * 41(3 ) * , 45 - 67 ( 1986 ) . wilson , g. : hilbert s sixteenth problem . topology , * 17(1 ) * , 53 - 74 ( 1978 ) .	a quadric in @xmath0 cuts a curve of degree 6 on a cubic surface in @xmath0 . the papers classifies the nonsingular curves cut in this way on non - singular cubic surfaces up to homeomorphism . two issues new in the study related to the first part of the 16th hilbert problem appear in this classification . one is the distribution of the components of the curve between the components of the non - connected cubic surface which turns out to depend on the patterns of arrangements ( see theorem 1 ) . the other is presence of positive genus among the components of the complement and genus - related restrictions ( see theorems 3 and 4 ) . = 6.15pt [ section ] [ thm]lemma [ thm]proposition [ thm]addendum [ thm]remark [ thm]example
the last cosmological and astrophysical data of large scale structure , the observations of type ia and cosmic microwave background radiation have demonstrated that currently there is an acceleration expansion phase in the universe @xcite . the cosmic expansion can be well described by a negative pressure so - called dark energy ( de ) . the simplest candidate for de is the cosmological constant . however , the cosmological constant suffers from the fine - tuning and the cosmic coincidence problems @xcite,@xcite,@xcite . therefore , to avoid these problems , different models for dark energy have been proposed such as quintessence , k - essence , tachyon @xcite , phantom , quintom @xcite , and the quantum gravity models , as well as holographic and new agegraphic models @xcite,@xcite . the tachyon model as a scalar field model arises in particle physics and string theory . thus , it can be considered as one of the potential candidates to describe the nature of the de . on the other hand , the problem of structure formation in the universe is a very important issue in theoretical cosmology . a simple model of structure formation is the spherical collapse model . the spherical collapse model was presented by gunn and gott @xcite . this model studies the evolution of growth of overdense structures with respect to the dynamics of scale factor or cosmic redshift . the dynamics of overdense structures depends on the dynamics of the background hubble flow and expansion of the universe . in the frame of general relativity , the spherical collapse model has been studied @xcite,@xcite,@xcite . in this paper , we study the spherical collapse and the evolution of spherical overdensities in the framework of tachyon scalar field model and compare the results with the results of einstein - de sitter ( eds ) and @xmath8-cold dark matter ( @xmath7 ) models . the action of tachyon scalar field over a cosmological background is given by @xcite @xmath9 where @xmath10 and @xmath11 are the tachyon scalar field and tachyon potential , respectively , and @xmath12 is the friedmann - robertson - walker ( frw ) metric having the cosmic time @xmath13 dependent scale factor @xmath14 . for a homogeneous field , the equation of motion is obtained as @xmath15 where the symbols @xmath16 and @xmath17 denote the derivatives with respect to @xmath13 and @xmath10 , respectively , and @xmath18 is called the hubble parameter . in the flat frw universe , the energy density @xmath19 and the pressure @xmath20 of the tachyon field read as @xmath21 @xmath22 for the pressureless matter and tachyon scalar field matter , the friedmann equation is given by @xmath23 where @xmath19 is the energy density of tachyon scalar field , and @xmath24 is the density of pressureless matter . we suppose that there is no interaction between @xmath19 and @xmath24 , so the continuity equations are given separately by @xmath25 @xmath26 using eqs . ( [ rho ] ) and ( [ p ] ) and also @xmath27 , the equation of state parameter ( eos ) for tachyon scalar field is obtained as @xmath28 the requirement for a real @xmath19 results in @xmath29 according to which @xmath30 should vary as @xmath31 . the fractional energy densities are defined by @xmath32 @xmath33 taking time derivative of eq . ( [ omega ] ) and using eq . ( [ dot ] ) yields @xmath34.\label{domega}\ ] ] also , taking time derivative of eq . ( [ friedman ] ) and using eqs . ( [ dot ] ) and ( [ dotm ] ) yields @xmath35 using eq . ( [ hdot ] ) and inserting eq . ( [ domega ] ) , we obtain @xmath36 here , the prime is the derivative with respect to @xmath37 where @xmath38 and z is the cosmic redshift . using @xmath39 and eq . ( [ eos ] ) , one finds @xmath40 the differential equation for the evolution of dimensionless hubble parameter , @xmath41 , in tachyon scalar field model , is obtained by using eqs . ( [ dot ] ) , ( [ dotm ] ) , ( [ eos ] ) and ( [ hdot ] ) as follows @xmath42.\label{ddf}\ ] ] the cosmological acceleration equation is also given by @xmath43 using eqs . ( [ rho ] ) and ( [ eos ] ) and inserting in eq . ( [ ddd ] ) , the acceleration equation reads as @xmath44 which can be converted to obtain the tachyon scalar field kinetic energy , using eqs . ( [ rho ] ) , ( [ friedman ] ) , and @xmath45 , as follows @xmath46.\label{phidot}\ ] ] with no loss of generality , we suppose the power law behavior @xmath47 and use @xmath48 to obtain @xmath49.\label{phi2}\ ] ] inserting ( [ phi2 ] ) in eqs . ( [ eos ] ) , ( [ df ] ) , ( [ ddf ] ) we can get the evolution of eos parameter @xmath50 , the density parameter of dark energy @xmath51 , and the dimensionless hubble parameter @xmath52 in tachyon scalar field model as a function of cosmic redshift . in figure ( 1 ) , assuming the present values @xmath53 , @xmath54 and @xmath55 , we have shown the evolution of eos parameter , the evolution of density parameter and the evolution of dimensionless hubble parameter of tachyon scalar field model with respect to the redshift parameter @xmath56 for @xmath57 . the evolution of dark energy density parameter of tachyon scalar field model with respect to the redshift parameter @xmath56 . , respectively . the dotted line shows the @xmath7 model.,title="fig:",width=192 ] the evolution of dimensionless hubble parameter in tachyon scalar field model and in the @xmath58 model . , respectively . the dotted line shows the @xmath7 model.,title="fig:",width=192 ] in this section , we study the linear growth of perturbation of non relativistic dust matter by computing the evolution of growth factor @xmath59 in tachyon scalar field model , and then compare it with the evolution of growth factor in eds and @xmath7 models . the differential equation for the evolution of the growth factor @xmath59 is given by @xcite,@xcite,@xcite @xmath60 in order to study the linear growth in tachyon scalar field model , using eqs.([df ] ) , ( [ ddf ] ) and ( [ phi2 ] ) , we solve numerically eq.([ga ] ) . also , we solve numerically eq.([ga ] ) for the eds model and the @xmath7 model . , respectively . the dotted line indicates the @xmath7 model and the dashed line represent the @xmath61 model . , title="fig:",width=192 ] in figure ( 2 ) , we have plotted the evolution of growth factor @xmath59 with respect to the scale factor . at first , namely for small scale factors , the growth factor in the tachyon scalar field model is larger than those of eds and @xmath7 models . however , for rather larger scale factors , the growth factor in the tachyon scalar field model becomes smaller than the eds model while it is still larger enough than that of @xmath7 model . this means that , at the beginning , the tachyon scalar field model predicts structure formation more efficient than eds and @xmath7 models . for later times , however , the structure formation in the tachyon scalar field model is dropped behind that of eds model , whereas it precedes the structure formation in the @xmath7 model . the structure formation is described by a non - linear differential equation for the evolution of the matter perturbation @xmath62 in a matter dominated universe @xcite,@xcite . in @xcite this differential equation was generalized to the case of evolution of @xmath62 in a universe including a dark energy component . now , we consider the non - linear differential equation which is given by @xcite @xmath63 where @xmath64 denotes the derivative with respect to @xmath65 . in the linear regime , we have @xmath66 in both cases , we consider the initial conditions @xmath67 , @xmath68 and @xmath69 @xcite . ( b)the non - linear growth of density perturbation@xmath62 in terms of@xmath65 . , respectively . the dotted line indicates the @xmath7 model and the dashed line indicates the eds model.,title="fig:",width=192 ] the figure ( 3-a ) shows that the linear growth factor in the tachyon scalar field model is larger than those of eds and the @xmath7 models , and the figure ( 3-b ) shows that the non - linear growth factor in the tachyon scalar field model is larger than the eds model . as time passes , the perturbation is growing and one can no longer use the linear regime . at this stage , the radius of perturbation region becomes maximal @xmath72 and the perturbation stops growing . this condition is called turn - around which points to the epoch when the grows of perturbation decouples from the hubble flow of the homogenous background . after the turn - around the perturbation starts contracting . for a perfect spherical symmetry and perfect pressureless matter , the perturbation would collapse to a single point becoming infinitely dense . since there is hardly any perfect spherical symmetric overdensity in the universe , the perturbation does not collapse to a single point and finally a virialized object of a certain finite size in equilibrium state is formed that is called halo . we call ( @xmath73 , @xmath74 ) and ( @xmath75 , @xmath76 ) as the redshift and radius corresponding to virialization and the turn - around epochs , respectively . now , we peruse two characterising quantities of the spherical collapse model for the tachyon scalar field model : the virial overdensity @xmath71 and the linear overdensity parameter @xmath70 . we consider a spherical overdense region with matter density @xmath77 in a surrounding universe described by its background dynamics and density @xmath78 . the virial overdensity @xmath71 is defined by @xcite @xmath79 which is a function of scale factor and redshift . we can rewrite the virial overdenty @xmath71 as follows @xcite @xmath80 where @xmath81 @xmath82 here , @xmath83 is the virial radius which is given by @xcite @xmath84 where @xmath85 and @xmath86 are the ( wang - steinhardt ) ws parameters @xmath87 @xmath88 now , we discuss the results obtained for the linear overdensity parameter and the virial overdensity for the models introduced in this work . the figure ( [ t2 ] ) shows the time evolution of linear overdensity , @xmath89 in terms of a function of the collapse redshift for the @xmath7 model , the eds model and the tachyon scalar field model . in the eds model , @xmath90 is independent of the redshift , hence it has a constant value i.e. @xmath91 . in the @xmath7 model , @xmath90 is smaller than 1.686 but the time evolution of the linear overdensity approaches the value of the eds model at high redshifts . , in terms of a function of the collapse redshift for the @xmath7 model , the eds model and the tachyon scalar field model . the blue , green and red lines represent the tachyon scalar field model for @xmath57 , respectively . the dotted line indicates the @xmath7 model and the dashed line indicates the eds model . , title="fig:",width=192 ] in fact , at high redshifts we have a matter dominated universe ( dust matter ) , but at lower redshifts we have a dark energy dominated universe , thus the structure formation must occur earlier . in the tachyon scalar field model , @xmath90 drives more slowly than the @xmath7 and the eds models because in fig . ( 1 ) , the hubble parameter in the tachyon scalar field model is larger than that of @xmath7 model . in terms of the collapse redshift @xmath73 for the @xmath7 model , the eds model and the tachyon scalar field model . the blue , green and red lines represent the tachyon scalar field model for @xmath57 , respectively . the dotted line indicates the @xmath7 model and the dashed line indicates the eds model . , title="fig:",width=192 ] in the figure ( [ u2 ] ) , we represent @xmath2 in terms of @xmath73 for the @xmath7 model , the eds model and the tachyon scalar field model . in the eds model , @xmath2 is independent of the redshift , thus it has a constant value i.e. @xmath92 . in the @xmath7 model , @xmath2 is smaller than 0.5 but it approaches the value of the eds model at high redshifts . in the tachyon scalar field model , @xmath2 drives more slowly than the @xmath7 and the eds models but its value approaches the value of the eds model at high redshifts . therefore , we can conclude that the size of structures in the @xmath7 model is larger than the tachyon scalar field model . \(b ) the variation of @xmath4 - @xmath73 for the @xmath7 model , the eds model and the tachyon scalar field model . , respectively . the dotted line indicates the @xmath7 model and the dashed line indicates the eds model . , title="fig:",width=192 ] in the figure ( 6-a ) , we represent @xmath3 in terms of @xmath73 for the @xmath7 model , the eds model and the tachyon scalar field model . in the eds model , @xmath3 is independent of redshift thus it has a constant value i.e. @xmath93 . in the @xmath7 model , @xmath3 is larger than 5.6 but its value approaches the value of the eds model in terms of high redshifts . in the tachyon scalar field model , @xmath3 drives faster than the @xmath7 model and the eds model but its value approaches the value of the eds model in terms of high redshifts . therefore , we can conclude that in the tachyon scalar field model the overdense spherical regions in terms of @xmath73 are denser than the @xmath7 model and the eds model . in the figure ( 6-b ) , we show the virial overdensity @xmath4 in terms of @xmath73 for the @xmath7 model , the eds model and the tachyon scalar field model . in the eds model , @xmath4 is independent of redshift , thus it has a constant value , @xmath94 . in the @xmath7 model , @xmath4 is more slowly than 178 , but its value approaches the value of the eds model in terms of high redshifts . in the tachyon scalar field model , @xmath4 drives faster than the @xmath7 model and its value approaches the value of the eds model in terms of high redshifts . the evolution of virial overdensity parameter @xmath4 is the main quantity for the halo size . therefore , we can conclude that in the tachyon scalar field model the halo size is larger than the eds model and the @xmath7 model . the average comoving number density of halos of mass m is given by @xcite,@xcite @xmath95 where @xmath96 and @xmath77 are the multiplicity function and the background density , respectively and @xmath97 is given by @xmath98 here @xmath99 is the r.m.s of the mass fluctuation in sphere of mass m. we can use the formula given by @xcite @xmath100 where @xmath101 is the mass variance of the overdensity on the scale of @xmath102 , @xmath103 and @xmath104 are the mass and the radius inside a sphere . also , @xmath105 is given by @xmath106 where @xmath107 is the linear growth factor , @xmath108 and @xmath109,\label{2}\ ] ] where @xmath110 . ( [ 1 ] ) and ( [ 2 ] ) have a validation range @xcite . they express that the fitting formula predicts higher values of the variance for @xmath111 and the fitting formula predicts lower values of the variance for @xmath112 . following @xcite , we apply st mass function @xmath113\exp(-\frac{b\nu}{2}).\label{st}\ ] ] here @xmath114 , @xmath115 and @xmath116 . we use the mass function introduced by del popolo ( po mass function)@xcite @xmath117\nonumber\\ \exp\big[-0.4019 b\nu\big(1+\frac{0.5526}{(b\nu)^{0.585}}+\frac{0.02}{(b\nu)^{0.4}}\big)^{2}\big],\end{aligned}\ ] ] where @xmath118 . also , we use the mass function ( yny mass function ) presented in @xcite @xmath119\nu^{(\frac{d}{2 } ) } \exp \big[-(b\sqrt{\frac{\nu}{2}})^{2}\big].\label{y}\ ] ] here @xmath120 , @xmath121 , @xmath122 and @xmath123 . now , we represent the evolution of the st mass function with respect to @xmath124 ( @xmath125 in figure ( 7 ) for the tachyon scalar field model and the @xmath7 model . we can see that the evolution of st mass function with respect to @xmath124 is the same for tachyon scalar field and the @xmath7 models in the @xmath126 case , but it is different for tachyon scalar field and @xmath7 models in the @xmath127 case . ( b)the evolution of the mass function with respect to @xmath124 for tachyon scalar field model and @xmath7 model in the case @xmath127 . , respectively . the dotted line indicates the @xmath7 model . , title="fig:",width=192 ] now , using eqs . ( [ n ] ) and ( [ st ] ) , for the tachyon scalar field model and the @xmath7 model , we obtain the average comoving number density of halos of mass @xmath128 in the cases @xmath129 . in the figure ( [ jop ] ) , we can see explicitly the differences for the cases @xmath126 and @xmath127 . we can see that difference of the number densities of halo objects is negligible for small objects in the case @xmath127 . therefore , we can obtain the number density of halo objects for high mass , and we find that the number of objects per unit mass is increasing for high mass in the tachyon scalar field model . the evolution @xmath130$]- @xmath124 for the tachyon scalar field model and the @xmath7 model in the case @xmath127 . , respectively . the dotted line indicates the @xmath7 model . , title="fig:",width=192 ] also , using eqs . ( [ st ] ) , ( [ del ] ) and ( [ y ] ) , we can compare the various mass functions at @xmath131 in figure ( [ jop11 ] ) . we can see that the po mass function is larger than st mass function and yny mass function for all mass scales . in this paper , we have studied the evolution of spherical overdensities in tachyon scalar field model by assuming the scale factor behavior as @xmath0 . we have shown the evolution of the eos parameter , the evolution of the density parameter and the evolution of the dimensionless hubble parameter of tachyon scalar field model with respect to a function of @xmath56 for @xmath57 . we saw that at early times of the scale factor evolution , the growth factor in the tachyon scalar field model is faster than the eds and @xmath7 models . so , we concluded that the tachyon scalar field model makes the structure formation to occur sooner , in comparison to other models . at later times , however , we showed that the growth factor in the tachyon scalar field model is more slower than that of eds model . also , in the eds model , @xmath90 is independent of the redshift and thus it has a constant value @xmath91 . in the @xmath7 model , @xmath90 is smaller than 1.686 , but the time evolution of the linear overdensity approaches the value of eds model at high redshifts . in fact , at high redshifts , we have a matter dominated universe ( dust matter ) , but at lower redshifts we have a dark energy dominated universe , thus the structure formation occurs earlier . in the tachyon scalar field model , @xmath90 is driven more slower than that of @xmath7 and eds models , because in figure ( 1 ) , the hubble parameter in the tachyon scalar field model is larger than that of @xmath7 model . also , we have shown that in the eds model , @xmath2 is independent of the redshift , thus it has a constant value i.e. @xmath92 . moreover , the size of structures in the @xmath7 model was larger than that of tachyon scalar field model . in the eds model , @xmath3 is independent of the redshift , hence it has a constant value i.e. @xmath93 . we saw that in tachyon scalar field model , the overdense spherical regions with respect to @xmath73 were denser than those of @xmath7 and eds models . in the eds model , @xmath4 is independent of the redshift , so it has a constant value , @xmath94 . the evolution of virial overdensity parameter @xmath4 is the main quantity for the halo size . therefore , we find that in the tachyon scalar field model the halo size is larger than the eds and @xmath7 models . finally , we saw that the evolution of the st mass function with respect to @xmath124 is the same for tachyon scalar field and @xmath7 models in the @xmath126 case , but it is not the same for tachyon scalar field model and the @xmath7 model in the @xmath127 case . also , the evolution of the number density with respect to @xmath124 is the same for the tachyon scalar field and @xmath7 models in the @xmath126 case , but its evolution is not the same for the tachyon scalar field and @xmath7 models in the @xmath127 case . the difference of number densities of halo objects is negligible for small objects in the case @xmath127 case . therefore , in obtaining the number density of halo objects for high mass , we find that the number of objects per unit mass is increasing for high mass in the tachyon scalar field model . we would like to thank m. malekjani for giving us useful comments that helped us to improve the scientific content of the manuscript . l. r. , batista . r. c. , liberato . l. , rosenfeld . r. , 2007 , jcapp , 11 , 12 . r. j. , cole . s. , efstathiou . g. , kaiser . n. 1991 , apj , 379 , 440 . f. 1994 , apj , 433 , 1 . 2000 , nature 404 , 955 ; + perlmutter . s. , et al . 2003 , astrophys . j. 598 , 102 ; + seljak . u. , et al . 2005 , phys . rev . d 71 , 103515 . y. f. , saridakis . e. n. , setare . m. r. , xia . j. q. , 2010 , phys . e. j. , sami . m. , tsujikawa . t. s. 2006 , ijmpd , 15 , 1753 . m. , robbers . g. 2006 , jcapp , 6 , 26 ; + wetterich . c. 2004 , phys . b , 594 , 17 . a. j. , goldreich . p. 1984 , apj , 281 , 1 . g. w. 2002 , phys . b 537 , 1 . gunn . j. e. , gott . r. j. 1972 , apj , 176,1 . y. , shaham . j. 1985 , apj . s. , pace . f. , bartelmann . m. 2012 , phys.rev.d , 86 , 103002 . f. , waizmann . j. c. , bartelmann . m. , 2010 , mnras , 406 , 1865 . f. , moscardini . l , crittenden . r. , bartelmann . m. , pettorino . v. 2014 , mnras , 437 , 547 . t. 1996 , cosmology and astrophysics through problems , cambridge university press . t. 2003 , phys . 380 , 235 . press . h. w. , schechter . p. 1974 , apj , 187 , 425 . w. j. 2005 , a. a , 443 , 819 . 2006 , apj , 637 , 12 ; + popolo . 2006 , a. a , 448 , 439 . a. g. , et al . 1998 , astron . j. 116 , 1009 ; + perlmutter . s. , et al . 1999 , astrophys . j. 517 , 565 . s. b. , gunn . e. j. 1987 , apj . m. r. 2007 , phys . b 653,116 ; + setare . m. r. , sadeghi . j. , amani . a. r. , 2009 , phys . b 673 , 241 . sheth . r. k. , tormen . g. , 1999 , mnras , 308 , 119 ; + sheth . r. k. , tormen . g. 2002 , mnras , 329 , 61 . p. t. p. , liddle . a. r. , 1996 , mnras , 281 , 323 . l. , steinhardt . p. j. 1998 , astro . j. 508 , 483 . s. 1989 , reviews of modern physics , 61 , 1 . h. , nagashima . m. , yoshii . y. 2004 , apj , 605 , 709 .	we study the tachyon scalar field model in flat frw cosmology with the scale factor behavior @xmath0 . we consider the spherical collapse model and investigate the effects of the tachyon scalar field on the structure formation in flat frw universe . we calculate @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 $ ] and @xmath6 $ ] for the tachyon scalar field model and compare the results with the results of eds model and @xmath7 model . [ firstpage ] tachyon scalar field ; spherical over - densities .
the future space mission planck will be able to measure , with unprecedented angular resolution and sensitivity , the cosmic microwave background ( cmb ) anisotropy and polarization at 9 frequencies in the range 30857 ghz . in combination with balloon borne experiments such as boomerang , maxima-1 , tophat and the recently launched space mission map , these observational data will provide a unique base for investigation of the history and the large scale structure formation of the universe . the accuracy of the cosmological parameter extraction planned for the planck mission is determined by the corresponding accuracy of the systematic effects . systematic errors can be one of the most important sources of errors for high multipole range of the @xmath0 power spectrum ( @xcite ) . it is well known that extraction of the information about cosmological parameters such as baryonic density @xmath1 , cold dark matter density @xmath2 , hubble constant @xmath3 , and so on , needs additional information about the statistical characteristics of the measured cmb anisotropy signal from the sky . the pure cmb signal is assumed to be a realization of a random gaussian signal on the sphere with power spectrum @xmath0 . the gaussianity of the cmb signal means that all its statistical properties are specified by its power spectrum @xmath0 , which depend on @xmath4 and not on the phases . in the framework of the cmb observations the signal measured by different instruments at different frequencies , however , displays some peculiarities in observational as well as in foreground manifestations . this is why a variety of the methods of the correct information extraction from the cmb data sets are now under discussion . all these methods are somewhat complementary to each other in the future highly sensitive cmb experiments , due to different sensitivity of the methods to different characteristics of the signal . from a theoretical point of view , the power spectrum of the true cmb signal is independent of fourier rings , meaning that it does not depend on the azimuthal number @xmath5 . for a flat patch of the sky it corresponds to homogeneity and isotropy of the signal , without angular dependency of the power spectrum @xmath6 on @xmath7 , where @xmath8 . in reality , the cmb signal from the sky has a more complicated structure reflecting some artifacts of the observation and different kinds of foreground contaminations , which can destroy the isotropy of the power spectrum . we will focus on a few important sources causing artificial anisotropy of the map : ( i ) `` non - gaussianity '' ( inhomogeneity and anisotropy ) of the foregrounds in the map ; ( ii ) asymmetry of the beam shape , which is now the standard part of investigation on systematic effects ; ( iii ) correlations of the instrumental ( pixel ) noise ; ( iv ) low multipole modes , e.g. @xmath9 for the whole sky ( @xmath10 , where @xmath11 is the linear size of the path of the sky ) , which are statistically peculiar . some of the above mentioned sources of the @xmath0 anisotropies are frequency dependent , thus their contributions to the maps at different frequency channels of the planck are different . for example , the foregrounds such as dust emission , synchrotron , free free , as well as bright and faint point sources , have different frequency dependencies and intensities at 33 and 857 ghz range , which definitely can manifest as some sources of errors in the pixel pixel window function and in the corresponding correlation function of the signals . the influence of the low multipole modes ( point ( iv ) ) on the possible anisotropies of the maps in the flat sky approximation can be detected directly from the corresponding @xmath6 amplitudes of the power spectrum . this paper is mainly devoted to illustration of the idea about manifestation and estimation of the asymmetric ( elliptical or more irregular ) beam shapes incorporated in the pixelized data , using both analysis of the two - dimensional spectrum and of phases of the map ( @xcite ) . we concentrate on beam asymmetry estimation using simulated cmb map , which reflects directly the specific of the planck scan strategy , map making and noise level . there are some important issues related to the beam profiles of the antenna for the planck low frequency instrument ( lfi ) and high frequency instrument ( hfi ) frequency channels ( @xcite ) . for instance , down to the level -10 db at lfi , the antenna shapes have approximately elliptical forms and peculiarities will only be included at higher multipoles if the level decreases down to -20 db or less , according to the design of the focal plane unit ( fpu ) . thus , roughly speaking , decentralization of the feed horns in the fpu produces the optical distortions of the beam shapes from the circular gaussian shapes(@xcite ) . beam shape influence on the accuracy of the cmb anisotropy @xmath0 extraction from the observational data is related to the scanning strategy and pixelization of the maps from the time - ordered data(tod)(@xcite ) . during scanning of the cmb sky the antenna beam moves across the sky , meaning that antenna beam is a function of time . after pixelization of the tod the position of each pixel in the cmb map is related directly with some points in the time stream for which we need to obtain the information of the orientation of the beam and location of the beam center relative to each pixel . in principle , given the scanning strategy ( which for the planck mission is under discussion ) and the beam shapes for each frequency channel , we would be able to model the geometrical properties of the pixel beam shapes and their manifestation in the pixel pixel window functions incorporated in the cmb power spectrum @xmath0 . however , the computational cost would increase dramatically due to the complicated character of the pixel pixel beam matrix(@xcite ) . moreover , the scanning strategy and the instrumental noise combined with the systematic effects could transform the actual beam shape during the time of observation . we then should find some peculiarities of the complicated beam shape influence on the cmb signal . if the response of an antenna on the measured signal is linear and the cmb signal and the instrumental noise are gaussian and not correlated , then the information about the beam anisotropy obtaining by both methods ( power spectrum analysis and phase analysis ) is the same . however in the general case the sets of encoded information obtained by both methods are different . thus using both methods is desirable . the plan of this paper is as follows . in section 2 we give some definitions of cmb signals and discuss the basic model of the planck sky map . in section 3 we introduce a general power spectrum and phase analysis of cmb signal and the concept of beam shape extraction . in section 4 we describe the main idea and its analytical approach . the numerical results are presented in section 5 and the conclusion in section 6 . let us introduce the standard model of cmb experiment where tod contain the information about the signal ( and noise ) from a large numbers of the circular scans . we suppose for simplicity that all systematic errors are removed after a preliminary `` cleaning '' of the scans . in the temporal domain the observed signal @xmath12 is the combination of the cmb + foreground signal @xmath13 and random instrumental noise @xmath14 , @xmath15 where @xmath16 with @xmath17 being the multipole expansion of the time stream beam @xmath18 . in eq . ( [ eq : eq2 ] ) @xmath19 is the corresponding multipole coefficient of the cmb + foreground signal expansion on the sphere and @xmath20 is the spherical harmonics . following tegmark ( 1996 ) we will assume that map making algorithm is linear . the signal in each pixel @xmath21 is then @xmath22 where @xmath23 is the corresponding pointing matrix and @xmath21 represents the cmb plus foregrounds signals from the sky convolved by the pixel beam @xmath24 , @xmath25 @xmath26 is the instrumental noise in each pixel : @xmath27 and @xmath28 is the coefficient of the pixel noise expansion . @xmath29 is a two dimensional vector with the components ( @xmath30 ) to denote the location on the surface of the sphere . the pointing matrix @xmath31 depends on the scanning strategy of the observation . below , as a basic model , we will use the model of the planck mission scanning strategy discussed by burigana et al . ( 2000 ) with stable orientation of the spin axis in the ecliptic plane and without precession of the spin axis . any precession of the spin axis will cause additional ( regular ) spin axis modulation . instead of the fast rotating beam model described by wu et al . ( 2001 ) , we will use a stable orientationed beam model during sky crossing for the spin and optical axes , i.e. , during the rotation of the optical axis around the spin axis of the satellite , the orientation of the beam is stable with respect to the optical axis . in addition we will also assume that instrumental noise is non - correlated for a single time ordered scan , and between scans as well . definitely this model of instrumental noise is primitive , and needs modifications and more detailed investigation , but it nevertheless reflects , as shown in the next section , the geometrical properties of asymmetry of the beam and their manifestation in the pixelized maps by a given scanning strategy . under the assumptions mentioned above we will use the elliptical beam shape model of burigana et al . ( 2000 ) . for a small part of the flat sky approximation , we will use the cartesian coordinate system with the @xmath32 axis parallel to the scanning direction and @xmath33 axis perpendicular to it . we denote by @xmath34 and @xmath35 the position of the center of the beam at the moment @xmath36 . then the beam shape can be written as @xmath37 , \label{eq : eq5}\ ] ] with @xmath38 where @xmath39 is the rotation matrix which describes the orientation of the elliptical beam , @xmath40 with @xmath41 the angle between @xmath32 axis and the principal axis of the ellipse . the @xmath42-matrix denotes the beam dispersion along the ellipse principal axis , expressed as @xmath43 as mentioned by wu et al . ( 2001 ) , the signal in the time stream is not the pixel temperature in @xmath21 itself . the observed signal in each pixel @xmath44 depends on the orientation of the pixel beam @xmath45 and the location of its center . for a non - symmetric spatially dependent beam , the convolution of the signal with asymmetric beam immediately produces asymmetry coupled with the underlying signal , which affects the estimation of angular power spectrum . for beam shape above -20 db level , where the beam shape is elliptical , we will use the flat sky approximation in order to demonstrate how we can estimate the beam shape in the phases diagram . in such an approximation we can describe the signal and instrumental noise on some small area of the sky using fft method instead of time consuming spherical harmonic expansion . in addition we will use the stable pixelized beam orientation model for a small area of the sky to reflect the scanning strategy of the observation . under the assumption of complete sky coverage , the spherical harmonics @xmath46 can be expressed as a product of the legendre polynomials @xmath47 and the signal can be written as ( burigana et al , 1998 ) @xmath48 , \label{eq : eq52}\end{aligned}\ ] ] where @xmath49 and @xmath50 we assume for simplicity the cobe - like cubic pixelization , which satisfies the following symmetry properties : if @xmath51 , then @xmath52 also , and if @xmath53 , then @xmath54 . using eq . ( [ eq : eq52 ] ) we can introduce the phases @xmath55 of the signal on the map , by @xmath56 where @xmath57 $ ] is the phase of the signal from the sky and @xmath58 $ ] is the phase of the noise . provided that cmb anisotropy is a random gaussian field , the @xmath19 of eq . [ eq : eq2 ] coefficient is a random variable with zero mean ( @xmath59 ) and variance @xmath60 where @xmath61 is the standard kronecker symbol and @xmath0 the power spectrum . actually , the realization of the random cmb signal on the sphere is unique , which means that in a @xmath62 distribution on the sky we have a single realization of the phases only . we denote the combined signal by @xmath63 : @xmath64 and the power spectrum of this unique realization by @xmath65 . our task is to extract the information about the beam shape either using @xmath55 , @xmath65 or their combination . the manifestation of the beam shape asymmetry in the @xmath65 can be demonstrated in the following way . let us consider the following function @xmath66 qualitatively the properties of the function @xmath67 are the following . for small @xmath4 the cmb signals surpass the noise and the value of @xmath65 is determined by the cmb signal , so for this region @xmath68 . for larger @xmath4 , @xmath65 becomes smaller . because of the cosmic variance ( fluctuations of @xmath65 as a spectrum of the realization of a random process ) , some @xmath65 can become close to zero , @xmath69 where @xmath70 is a constant and @xmath71 . this means that @xmath67 at these points on the @xmath72-plane has maxima @xmath73 the number density of these maxima increases at larger @xmath4 . for @xmath4 of the order @xmath74 , where @xmath75 fwhm of the antenna beam , the antenna affects the spectrum . if the antenna is asymmetric , this influence is also asymmetric . thus we have the asymmetric distribution of maxima in this region of the spectral plane . for very large @xmath4 s , where the noise @xmath28 dominates the number density of maxima , eq . ( [ eq : delta2 ] ) is determined by the noise , and does not depend on the beam and is therefore isotropic . the effects are demonstrated in fig . [ deltapower ] , which is a result of a numerical experiment . in order to show the beam effect from @xmath76 , we add on the symmetric part . there are 25 contour levels between @xmath77 and @xmath78 . the coordinates represent the flat sky approximation of the general case at the limit @xmath79 ( described in section 5 ) . interval of modes that are sensitive to the beam asymmetry starts from @xmath80 and goes to infinity in ideal conditions ( no pixel noise , @xmath81 ) . for the real situation the limit of this interval is finite and determined by the pixel noise , @xmath82 . to extract the information about the beam shape from fig . [ deltapower ] one needs , for example , to draw the averaged iso - density of the distribution of maxima of @xmath83 satisfied to eq . ( [ eq : delta2 ] ) . one of the possible methods of drawing this is described in the next section . in this section we will show how to estimate the antenna beam shape using the information contained in the phase distribution of the signal in the map , using a single realization of the phases of all @xmath84 modes . after the description of this phase method it will be clear that , in the simple case of linear response of the antenna , gaussian cmb signal and noise , the phase method is equivalent to the power spectrum method described above . however , as we mentioned in the introduction , in the general case these methods give different sets of information . for the phase analysis we will `` perturb '' the phases by adding controlled white noise into the map . we consider an ensemble @xmath85 ( @xmath86 ) , where each element of the ensemble consists of the same realization of the cmb signal and pixel noise plus a random realization of a controlled white noise @xmath87 with the variance @xmath88 and random phases for each @xmath89 mode , @xmath90 we calculate the ensemble average of the squared difference between phases @xmath91 of the noise - added realization and the initial one , @xmath55 : @xmath92 the function @xmath93 from eq . ( [ eq : squareddifference ] ) is considered separately for different values of the variance of the controlled noise @xmath94 in the range from @xmath95 , up to the level of pixel noise @xmath96 . we will show that the function @xmath93 reflects all asymmetric peculiarities of the initial signal @xmath21 . the results of such a kind of phase analysis are tested numerically and are presented in the next section . here we describe the analytical approach to the analysis of the phase mixing effect to demonstrate how it is possible to reconstruct the antenna beam shape from the phase distributions in @xmath89-plane . for the @xmath97 signal the definition of the combined phases @xmath98 for each @xmath89 mode is similar to eq . ( [ eq : tangentequation ] ) : @xmath99 where @xmath63 is the multipole expansion of the combined signal(cmb @xmath100 beam + pixel noise ) at the mode @xmath89 , @xmath55 the corresponding phase , @xmath101 the controlled noise expansion , and @xmath102 its phase . the analytical expression for the function @xmath93 can be written in the following way @xmath103 \right\ } , \label{eq : delta4}\end{aligned}\ ] ] where @xmath104 and @xmath105 the integral in eq . ( [ eq : delta4 ] ) has been tabulated as a function of two variables @xmath55 and @xmath106 and the result is presented in fig . [ tabulate ] . we will use this tabulation for the numerical experiment in the next section . in this section we will obtain simple analytical asymptotics of eq . ( [ eq : squareddifference ] ) . from eq . ( [ eq : delta4 ] ) one can find the difference @xmath107 let us determine the mean squared value @xmath108 of the difference between @xmath109 and @xmath110 using eq . ( [ eq : tangentdifference ] ) @xmath111 one finds that @xmath112 if @xmath113 and after integration we obtain @xmath114 \label{eq : meansqtangentsol}\end{aligned}\ ] ] where @xmath115 , and @xmath116 in the asymptotic @xmath117 we have @xmath118 and from eq . ( [ eq : meansquaredtangent ] ) and eq . ( [ eq : meansqtangentsol ] ) we get @xmath119 . \label{eq : approxmeansquared}\ ] ] for our approximation we can neglect the second term in the brackets of eq . ( [ eq : approxmeansquared ] ) and @xmath120 properties of @xmath121 are the same for any values of @xmath4 ( and @xmath5 ) . properties of @xmath122 were discussed in the previous section when we discussed @xmath67 . it is obvious that qualitatively the behaviors of the function @xmath93 are the same as @xmath67 . but there is one principal difference between the qualitative and the full correct description of the asymptotic @xmath123 . namely , if @xmath123 then @xmath124 , and the asymptotic eq . ( [ eq : approxmeansquared ] ) is not valid . under the condition @xmath125 we have @xmath126 and from eq . ( [ eq : squareddifference ] ) and eq . ( [ eq : tangent2equation ] ) we reach @xmath127 in fact the controlled white noise is some kind of averaging factor in our phase analysis . an analogous method may be used for the power spectrum analysis of the beam asymmetry . the general properties of @xmath128 allow us to introduce more convenient and faster flat sky analysis , which is specifically useful for the antenna beam shape estimation . this model reflects the scanning strategy at present specified for the planck mission , when , for a small part of the sky far from the north and south poles , the model of the stable and fixed beam orientation is adequate , without rotation and multi - crossing scans . the implementation of fft significantly decreases the computational cost , which , for time consuming spherical harmonic analysis , is a major issue . using definitions of the signals and noises from the previous section , we define @xmath129 to be the fourier component of the signal from the sky measured by antenna and @xmath130 that of the noise , which does not depend on the beam properties . the fourier modulus and phases of the combined signal are defined as follows , @xmath131 . \label{eq : eqf1}\ ] ] the function @xmath6 is the power spectrum of the combined signal ( observational data and noise ) and @xmath132 describes the phases . @xmath129 represents the cmb signal convolved with the beam . if the signal from the sky does not contain non - gaussian foreground components ( or its manifestation is suppressed at the `` cosmologically clean '' channels from 70 to 217 ghz ( @xcite ) ) and if the noise @xmath130 is uncorrelated white noise , the distribution of the phases @xmath132 for different @xmath133 must be random and uniform . if there are some contaminations from the foregrounds and(or ) pixel noise is of non - gaussian nature , the phase distribution over the @xmath133 range can be weakly correlated , and can be tested by phase diagram method(@xcite ) . such kind of uncertainties of the statistical properties of the signal @xmath6 are important for separation of the pure cmb signal and noise for complicated characters of the beam shape . we start out with a squared gaussian random map with the power spectrum from the angular power spectrum of the @xmath134cdm model from lee et al . the map simulates a @xmath135 square realization of the cmb temperature fluctuations with pixel size 3 arcmin and periodic boundary conditions ( pbc ) ( see @xcite and remarks in section 6 ) . we then add pixel noise , @xmath136 , after convolving the map with an elliptical beam with long - axis and short - axis fwhm 12 and 9 arcmin , respectively . after generation of the map , which models the hfi 100 ghz frequency channel , we sum @xmath137 realizations of the controlled noise with variance @xmath138 close to the pixel noise variance ( eq . ( [ eq : addwhitenoise ] ) and @xmath139 ) and calculate the mean squared difference between phase of the signal and that of the signal plus controlled noise in the flat sky approximation : @xmath140 ^ 2\rangle \|_m . \label{eq : eq99}\ ] ] analogously we can calculate @xmath141 for the flat approximation . in fig . [ deltapower ] we display the function @xmath141 , it is rather difficult , however , to draw the averaged contour lines for @xmath141 directly from this map . it is easier to do this for the map of the @xmath142 function . [ colorphase ] displays the @xmath143 function using color representation of phases ( @xcite ) . in the color representation , phases are mapped onto the color circle , so that the phase of each mode is represented by a color hue . color red represents @xmath144 and the primary hues are @xmath145 apart , i.e. , phases of @xmath145 and @xmath146 are represented by color green and blue . cyan and other complementary colors magenta and yellow represents @xmath147 and @xmath148 , and @xmath149 , respectively . for presentation effect , we add the symmetric part ( fourier ring ) in the fourier domain . it is clear from the color map of fig . [ colorphase ] that in the center for large - scale modes ( small @xmath150 ) , the values of @xmath143 are approaching zero , shown by color red . this is due to the fact that the phases of large - scale modes ( the inner rings ) are not `` perturbed '' by added noise . on small - scale modes ( the outer rings ) , where the phases are dominated by pixel noise , the added controlled noise randomizes the resulting phases when the controlled noise level is chosen the same as that of pixel noise . in the intermediate regime , the beam shape manifests itself through the @xmath143 function . the fuzzy regions in the intermediate regime reflect the anisotropy of the beam shape . by taking average as @xmath151 where @xmath152 , @xmath153 , we can find the mean shape of the beam . figure [ contour ] shows the contour map of the @xmath154 function by @xmath155 . from the contour map , the elliptic shape is estimated roughly to have the same ratio of the simulated beam shape , i.e. , 4/3 . an important issue is related to the asymmetric beam extraction from a real map , covering a small patch of the sky . previously we used periodic boundary condition ( pbc ) for modeling the cmb signal ( @xcite ) . in reality , for some square @xmath156 patch of the whole sky map , the pbc is artificial and one can ask how sensitive this controlled noise method is to the deviation of the artificial phases of the signal from the true distribution ? to answer this question we show in fig . [ colorphasenp ] the result of a numerical experiment for a map which was constructed in the following way : we generate , as described earlier , the pbc map with the size @xmath157 , @xmath158 , and extracted from this map the inner part @xmath159 with the size @xmath160 . we then apply the controlled noise method for the non - pbc map and compare our results of the beam extraction with the pbc case . figure [ colorphasenp ] shows the colored map of this @xmath161 function . in order to compare with the pbc case , only half of @xmath162-range of interest is extracted from non - pbc case , as the beam size relative to the map is now twice of the pbc case . the peculiarity of the red crossing is induced by the non - pbc . the difference between these two models is less then @xmath163 . this implies that we can apply our method for small real patch of the sky directly . in this paper we propose the method of power and phase analysis to demonstrate the manifestation of the anisotropy of the antenna beam shape incorporated in the pixelized cmb anisotropy data , and how to estimate the shape of the beam . it is worth noting that the numerical realization of the phase diagram method described above is based on the flat sky approximation and illustrates the general properties of the cmb signal extraction from the pixelized sky map at the high multipole limit @xmath164 . in such a case , the fourier analysis reflects directly the general properties of the phases of a signal on the sphere , which in the general ( whole sky ) limit must be defined using spherical harmonic analysis . but there are a few balloon experiments ( boomerang , maxima , tophat ) which cover relatively small areas of the sky in comparison with planck and the recently launched map missions . moreover , for the beam shape extraction from map and planck missions it will be convenient to extract preliminary information about antenna beam shape without time consuming spherical harmonic computations . in connection with the planck mission this approach looks promising due to otherwise high computation cost in the framework of the @xmath0 extraction program . as it is shown , the functions @xmath165 and @xmath166 reflect the general properties of the power spectrum @xmath167 measured from the small patch of the sky . at the end of this discussion we would like to give the following remark . for estimation of the asymmetry of the antenna beam shape using the controlled noise method , we need to know the limit of the beam cross - level ( in db ) , for which we can extract the peculiarities of the beam . according to general prediction of the beam shape properties for the planck mission it is realistic to assume that the ellipticity of the beam preserves up to the cross - level @xmath168 db . for a small patch of the map we can use @xmath169 $ ] and @xmath6 for the best fit @xmath134cdm cosmological model from the maxima-1 data . because of logarithmic dependence of the beam shape parameters on the @xmath6 and @xmath170 : @xmath171\ ] ] we can find the ratio @xmath172 with 3 - 5 @xmath173 accuracy at the cross - level @xmath174 db . this method is most applicable to the patch of sky at low declination where the assumption of stable orientation of the antenna beam is satisfied . a more general case will be investigated in a separate paper . this paper was supported in part by danmarks grundforskningsfond through its support for the establishment of the theoretical astrophysics center , by grants rfbr 17625 and intas 97 - 1192 .	we present a new method to extract the beam shape incorporated in the pixelized map of cmb experiments . this method is based on the interplay of the amplitudes and phases of the signal and instrumental noise . by adding controlled white noise onto the map , the phases are perturbed in such a way that the beam shape manifests itself through the mean - squared value of the difference between original and perturbed phases . this method is useful in extracting preliminary antenna beam shape without time - consuming spherical harmonic computations .
shape oscillations of levitated metal droplets can be used to measure the surface tension and viscosity of liquid metals ( @xcite ) . theoretically , the former determines the frequency , while the latter accounts for the damping rate of oscillations . in the reality , experimental measurements may be affected by several side - effects . firstly , levitated drops may be significantly aspherical and the oscillations amplitudes not necessarily small , whereas the classical theories describing the oscillation frequencies ( @xcite ) and damping rates ( @xcite ) assume small - amplitude oscillations about an ideally spherical equilibrium shape . corrections due to the drop asphericity have been calculated by @xcite and @xcite . @xcite find the same order correction to the damping rate resulting also from ac - magnetic field . the effect of a moderate amplitude on the oscillations of inviscid drops has been analysed by @xcite who find that the oscillation frequency decreases with the square of the amplitude . using a boundary - integral method , @xcite show that small viscosity has a relatively large effect on the resonant - mode coupling phenomena in the nonlinear oscillations of large axially symmetric drops in zero gravity . numerical simulation of large - amplitude axisymmetric oscillations of viscous liquid drop by @xcite , who uses the galerkin / finite - element technique , shows that a finite viscosity results in a much stronger mode coupling than predicted by the small - viscosity approximation . secondly , the measurements may strongly be disturbed by ac - driven flow in the drop . the mode coupling by the internal circulation in axisymmetrically oscillating drop has been studied numerically by @xcite using the galerkin / finite - element technique . to reduce the strength of the ac field necessary for the levitation and , thus , to minimise the flow , experiments may be conducted under the microgravity conditions during parabolic flights or on the board of space station ( @xcite ) . a cheaper alternative might be to apply a sufficiently strong dc magnetic field that can not only stabilise ac - driven flow but also suppress the convective heat and momentum transport responsible for the mode coupling under the terrestrial conditions as originally shown by @xcite . such an approach has been implemented first by @xcite on the electromagnetically levitated drops of copper and nickel which were submitted to a dc field of the induction up to @xmath0 the only motion of @xmath1 drops observed to persist in magnetic field above @xmath2 was a solid - body rotation about an axis parallel to the magnetic field . no shape oscillations , usually induced by the ac - driven flow fluctuations , were observed . note that this implies only the suppression of ac - driven flow but not of the shape oscillations themselves which require an external excitation to be observable . @xcite study the effect of suppression of the melt flow on the structure of various alloys obtained by the electromagnetic levitation melting technique in a strong superimposed dc magnetic field . the use of high magnetic fields in various material processing applications is reviewed by @xcite . note that a similar suppression of ac - driven flow can also be achieved by a fast spinning of the drop ( @xcite ) that may be driven by an electromagnetic spin - up instability ( @xcite ) . the effects of both the drop rotation and ac - driven flow on the frequency spectrum of shape oscillations have been modelled numerically by @xcite . @xcite demonstrates numerically that a large enough oscillation amplitude can compensate for the effect of rotation on the frequency shift . a novel method of measuring thermal conductivity of liquid silicon using the electromagnetic levitation in a strong superimposed dc magnetic has been introduced by @xcite . subsequent numerical modelling by @xcite shows that applying a dc magnetic field of @xmath3 can suppress convection in molten silicon droplet enough to measure its real thermal conductivity . later on this method has been extended to the measurements of heat capacity of molten austenitic stainless steel ( @xcite ) and also that of supercooled liquid silicon ( @xcite ) . in order to determine the surface tension and viscosity or the electrical conductivity one needs to relate the observed surface oscillations with the relevant thermophysical properties of the liquid . general small - amplitude shape oscillations of conducting drop in a uniform dc magnetic field have been analysed first by @xcite . although a magnetic field of arbitrary strength is considered , the solution is restricted to inviscid drops . moreover , only the frequency spectrum and magnetic damping rates are found but not the associated shape eigenmodes , which may be useful for experimental identification of the oscillation modes . energy dissipation by axisymmetric oscillations of a conducting drop in a weak dc magnetic field is considered by @xcite , who finds the magnetic damping rates in agreement with more general results of @xcite . axisymmetric oscillations of an electromagnetically levitated drop of molten @xmath4 in a superimposed dc magnetic field are modelled numerically by @xcite . a moderate dc magnetic field is shown to stabilise ac - driven flow and , thus , to eliminate the associated shape oscillations . a three - dimensional numerical simulation of an oscillating liquid metal drop in a uniform static magnetic field has been carried out by @xcite . the numerical results show that vertical magnetic field effectively damps the flow , while horizontal field tries to render the flow two - dimensional . in the present paper , we analyse free oscillations of a viscous electrically conducting drop in a homogeneous dc magnetic field . in contrast to @xcite , we assume the viscosity to be small but non - zero and the magnetic field to be strong . this allows us to obtain an asymptotic solution to the eigenvalue problem for general small - amplitude 3d shape oscillations including the eigenmodes left out by @xcite , which are necessary for the subsequent determination of the viscous damping . firstly , we show that the eigenmodes of shape oscillations are not affected by strong magnetic field . namely , they remain the spherical harmonics as in the non - magnetic case . the magnetic field , however , changes the internal flow associated with the surface oscillations and , thus , the frequency spectrum . as the drop oscillates in a strong magnetic field , the liquid moves as solid columns aligned with the field . two types of such oscillations are possible : longitudinal and transversal to the magnetic field . the oscillations are weakly damped by a strong magnetic field , except for both the axisymmetric transversal and inherently 2d modes . the former are magnetically overdamped because the incompressibility constraint does not permit an axially uniform radial flow . the latter , which are transversal modes defined by the spherical harmonics with equal degree and order , @xmath5 , are not affected at all because these modes are naturally invariant along the field . because the magnetic damping for all other modes decreases inversely with the square of the field strength , the viscous damping may become important in a sufficiently strong magnetic field . the paper is organised as follows . the problem is formulated in @xmath6[sec : prob ] . section [ sec : invisc ] presents an inviscid asymptotic solution which yields the shape eigenmodes and frequency spectrum of longitudinal and transversal oscillations . magnetic damping is found in @xmath6[sub : mag - damp ] as a next - order asymptotic correction to the frequency . viscous damping rates are calculated in @xmath6[sub : visc - damp ] first by the eigenvalue perturbation technique for the longitudinal modes and then by an energy dissipation approach for both of the oscillation modes . the paper is concluded by a summary and discussion of the results in @xmath6[sec : conc ] . sketch to the formulation of problem.,scaledwidth=33.0% ] consider a spherical non - magnetic drop of an incompressible liquid with radius @xmath7 density @xmath8 surface tension @xmath9 electrical conductivity @xmath10 and a small dynamic viscosity @xmath11 performing small - amplitude shape oscillations in a strong uniform dc magnetic field @xmath12 as illustrated in figure [ fig : sketch ] . the velocity of liquid flow @xmath13 and the pressure distribution @xmath14 are governed by the navier - stokes equation with electromagnetic body force@xmath15 where the induced current follows from ohm s law for a moving medium @xmath16 owing to the smallness of oscillation amplitude , the nonlinear term in ( [ eq : ns ] ) as well as the induced magnetic field are both negligible . in addition , the characteristic oscillation period @xmath17 is supposed to be much longer than the magnetic diffusion time @xmath18 where @xmath19 is the permeability of vacuum . this leads to the quasi - stationary approximation according to which @xmath20 and @xmath21 where @xmath22 is the electric potential . the incompressibility constraint @xmath23 and the charge conservation condition @xmath24 applied to ( [ eq : ns ] ) and ( [ eq : ohm ] ) result , respectively , in@xmath25 for a uniform @xmath12 under consideration here , applying the operators @xmath26 @xmath27 and @xmath28 to ( [ eq : ns ] ) and taking into account @xmath29 together with ( [ eq : p ] ) and ( [ eq : phi ] ) , we obtain@xmath30\{p,\varphi,\vec{v}\}=0.\label{eq : pfv}\ ] ] although the equation above applies to @xmath31 @xmath22 and @xmath13 separately , these variables are not independent of each other . firstly , owing to the incompressibility constraint , only two velocity components are mutually independent . secondly , velocity is related to the pressure and electric potential by ( [ eq : ns ] ) , which can be used to represent @xmath13 in terms of @xmath14 and @xmath22 as done in the following . boundary conditions are applied at the drop surface @xmath32 defined by its spherical radius @xmath33 where @xmath34 is a small perturbation , which depends on the poloidal and azimuthal angles , @xmath35 and @xmath36 and the time @xmath37 . the radial velocity at the surface is related to the radius perturbation by the kinematic constraint @xmath38 the normal component of the current at the drop surface , which is assumed to be surrounded by vacuum or insulating gas , vanishes , i.e. , @xmath39 in addition , there is no tangential stress at the free surface : @xmath40 while the normal stress component is balanced by the capillary pressure @xmath41 where @xmath42 is the constant part of pressure , @xmath43 is a unit tangent vector and @xmath44 is the outward surface normal . for small - amplitude oscillations defined by @xmath45 we have @xmath46 henceforth , we proceed to dimensionless variables by choosing the radius @xmath47 and the characteristic capillary pressure @xmath48 as the length and pressure scales . the characteristic period of capillary oscillations is determined by the balance of inertia and pressure , which yields the time scale @xmath49 the velocity and potential scales are chosen as @xmath50 and @xmath51 respectively , where @xmath52 in the dimensionless variables , ( [ eq : pfv ] ) takes the form@xmath53\{p,\varphi,\vec{v}\}=0,\label{eq : pfv - nd}\ ] ] where @xmath54 is a unit vector in the direction of the magnetic field and @xmath55 and @xmath56 are the conventional and magnetic capillary numbers , respectively . they are the ratios of the capillary oscillation time @xmath17 defined above and the viscous and magnetic damping times , which are @xmath57 and @xmath58 respectively . in the dimensionless form , the normal stress balance condition ( [ eq : nblnc ] ) reads as @xmath59 in the following , we assume viscosity to be small but the magnetic field strong so that @xmath60 and @xmath61 which means that the second and third terms in ( [ eq : pfv - nd ] ) are much greater and much smaller , respectively , than the first one . thus , we first focus on the effect of the magnetic field and ignore that of viscosity , which is considered later in @xmath6[sub : visc - damp ] . here we ignore viscosity that allows us to formulate the problem in terms of @xmath31 @xmath22 and @xmath62 projecting the dimensionless counterpart of ( [ eq : ns ] ) , which takes the form @xmath63,\label{eq : ns - nd}\ ] ] onto @xmath64 and @xmath65 and putting @xmath66 we obtain @xmath67,\label{eq : ns - r}\\ \partial_{t}v_{\shortparallel } & = & -\vec{\epsilon}\cdot\vec{\nabla}p,\label{eq : ns-\|\|}\end{aligned}\ ] ] where @xmath68 is the velocity component along the magnetic field . differentiating ( [ eq : ns - r ] ) with respect to @xmath37 and substituting @xmath69 from ( [ eq : ns-\|\| ] ) , we represent ( [ eq : kinc ] ) in terms of @xmath14 and @xmath22@xmath70_{r=1}.\label{eq : r1-cnd}\ ] ] velocity has to be eliminated also from the electric boundary condition given by the radial component of ohm s law@xmath71_{r=1}=0.\label{eq : jr}\ ] ] firstly , applying @xmath72 to ( [ eq : jr ] ) and then using ( [ eq : ns - nd ] ) , we obtain @xmath73_{r=1}=0.\label{eq : phi - cnd}\ ] ] in the inviscid approximation , ( [ eq : nblnc - nd ] ) reduces to @xmath74 in the following , besides the spherical coordinates @xmath75 we will be using also the cylindrical ones @xmath76 with the axis aligned along the magnetic field so that @xmath77 solution is sought in the normal mode form @xmath78 where @xmath79 @xmath80 and @xmath81 are axisymmetric amplitude distributions , @xmath82 is the azimuthal wave number , and @xmath83 is a generally complex temporal variation rate which has to be determined depending on @xmath84 @xmath85 and @xmath86 then boundary conditions ( [ eq : r1-cnd ] ) , ( [ eq : phi - cnd ] ) and ( [ eq : p - cnd ] ) for the oscillation amplitudes at @xmath87 take the form @xmath88 where @xmath89 is the angular part of the laplace operator in the spherical coordinates for the azimuthal mode @xmath82 written in terms of @xmath90 further , it is important to note that@xmath91 where @xmath92 the associated legendre function of degree @xmath93 and order @xmath84 is an eigenfunction of @xmath94 with eigenvalue @xmath95 ( @xcite ) . equation ( [ eq : pfv - nd ] ) for @xmath96 and @xmath97 can be written as @xmath98\{\hat{p},\hat{\varphi}\}=0,\label{eq : phip}\ ] ] where @xmath99 is the radial part of the laplace operator in the cylindrical coordinates for the azimuthal mode @xmath100 here we put @xmath66 suppose @xmath61 and search for an asymptotic solution in the terms of a small parameter @xmath101 as@xmath102 note that although ( [ eq : phip ] ) admits solutions with @xmath103 found by @xcite , such quickly relaxing modes can not be related with the surface deformations . from the physical point of view , drop is driven to its equilibrium shape by the surface tension , and the magnetic field can only oppose but not to accelerate the associated liquid flow . from the mathematical point of view , @xmath104 applied to ( [ eq : rh - cnd ] ) results in @xmath105 which means no surface deformation at the leading order in agreement with the previous physical arguments . consequently , these fast modes represent internal flow perturbations which are not relevant for the shape deformations under consideration here . at the leading order , ( [ eq : phip ] ) reduces to @xmath106 whose general solution is @xmath107 where the first pair of particular solutions are the functions of @xmath108 only , while the second pair is linear in @xmath109 but general in @xmath110 owing to the @xmath109-reflection symmetry of the problem these two types particular of solutions do not mix and , thus , they are subsequently considered separately . we refer to these solutions in accordance to their @xmath109-parity as even and odd ones using the indices @xmath111 and @xmath112 as shown below , the odd and even solutions describe longitudinal and transversal oscillation modes , respectively . for the odd solutions @xmath113 boundary condition ( [ eq : phih - cnd ] ) , which at the leading order reads as @xmath114 results in @xmath115 the two remaining boundary conditions ( [ eq : rh - cnd ] ) and ( [ eq : ph - cnd ] ) take the form@xmath116 eliminating the pressure term between the equations above , we obtain an eigenvalue problem in @xmath117 for @xmath118 @xmath119 which is easily solved by using ( [ eq : legendre ] ) as@xmath120 where @xmath121 is a small amplitude of oscillations and @xmath122 is an odd positive number . note that imaginary @xmath123 describes constant - amplitude harmonic oscillations with the circular frequency @xmath124 which differs from the corresponding non - magnetic result only by the factor of @xmath125 ( @xcite ) , and coincides with the result stated by @xcite . thus , strong magnetic field changes only the eigenfrequencies but not the eigenmodes of shape oscillations which , as without the magnetic field , are represented by separate spherical functions ( associated legendre functions with integer indices ) ( @xcite ) . similarly to the non - magnetic case , the frequency spectrum for odd modes is degenerate because it depends only on the degree @xmath93 but not on the order @xmath100 thus , for each @xmath126 there are @xmath127 $ ] odd modes with different @xmath100 taking into account that @xmath128 at the surface , the radial pressure distribution is obtained from ( [ eq : r0-o ] ) as @xmath129 according to ( [ eq : ns-\|\| ] ) , this pressure distribution is associated with the axial velocity component @xmath130 while two other velocity components transversal to the magnetic field are absent in the leading - order approximation . thus , the liquid effectively oscillates in solid columns along the magnetic field as illustrated in figure [ fig : odd ] for the first four longitudinal oscillation modes defined by the indices @xmath131 , @xmath132 , @xmath133and @xmath134 since such a flow does not cross the flux lines , the oscillations are not damped by the magnetic field in the leading - order approximation . shapes and the associated liquid oscillations in the @xmath135-plane parallel to the magnetic field for the first four longitudinal oscillation modes with indices ( @xmath136 ( a ) , @xmath132 ( b ) , @xmath137 ( c ) , and ( 4,1 ) ( d).,title="fig:",scaledwidth=25.0%](-100,0)(a)shapes and the associated liquid oscillations in the @xmath135-plane parallel to the magnetic field for the first four longitudinal oscillation modes with indices ( @xmath136 ( a ) , @xmath132 ( b ) , @xmath137 ( c ) , and ( 4,1 ) ( d).,title="fig:",scaledwidth=25.0%](-100,0)(b)shapes and the associated liquid oscillations in the @xmath135-plane parallel to the magnetic field for the first four longitudinal oscillation modes with indices ( @xmath136 ( a ) , @xmath132 ( b ) , @xmath137 ( c ) , and ( 4,1 ) ( d).,title="fig:",scaledwidth=25.0%](-100,0)(c)shapes and the associated liquid oscillations in the @xmath135-plane parallel to the magnetic field for the first four longitudinal oscillation modes with indices ( @xmath136 ( a ) , @xmath132 ( b ) , @xmath137 ( c ) , and ( 4,1 ) ( d).,title="fig:",scaledwidth=25.0%](-100,0)(d ) for the even solutions @xmath138 the leading - order boundary condition ( [ eq : phih - cnd ] ) is satisfied automatically . the two remaining conditions ( [ eq : rh - cnd ] ) and ( [ eq : ph - cnd ] ) then take the form@xmath139 in contrast to the longitudinal modes considered above , now we have two equations ( [ eq : r0-e ] ) and ( [ eq : p0-e ] ) but three unknowns . to solve this problem , we need to consider the first - order solution to ( [ eq : phip ] ) which now takes the form @xmath140 and yields @xmath141 then boundary condition ( [ eq : phih - cnd ] ) results in @xmath142 combining this with ( [ eq : r0-e ] ) and ( [ eq : p0-e ] ) and taking into account @xmath143 we obtain @xmath144\hat{r}_{0}^{e}=0.\label{egv : r0-e}\ ] ] using ( [ eq : legendre ] ) , we readily obtain @xmath145 where @xmath146 is a small oscillation amplitude and @xmath122 is an even non - negative number . the result above again agrees with the asymptotic solution given by @xcite . similarly to the odd solutions found in the previous section , even eigenmodes are represented by separate spherical functions , and the oscillations are not damped at the leading order . in contrast to the odd modes as well as to the non - magnetic case , the frequency spectrum ( [ sol : bt0-e ] ) is no longer degenerate and frequencies vary with the azimuthal wave number @xmath82 . in particular , there are two important results implied by ( [ sol : bt0-e ] ) . firstly , the oscillation frequency for the axisymmetric modes specified by @xmath147 is zero . this means that these modes are over - damped and do not oscillate at all . secondly , the oscillation frequency for the modes with @xmath148 is exactly the same as without the magnetic field , i.e. , @xmath149 this is because the liquid flow associated with these oscillation modes is inherently invariant along the field and , thus , not affected by the last ( @xcite ) . the electric potential and pressure distributions follow from ( [ eq : r0-e ] ) and ( [ eq : p0-e ] ) as @xmath150 the associated velocity distribution is obtained from ( [ eq : ns - nd ] ) . firstly , equation ( [ eq : ns-\|\| ] ) implies that the liquid oscillations are purely transversal to the magnetic field . in the leading - order terms , we obtain from ( [ eq : ns - nd ] ) @xmath151 which shows that the velocity is not only transversal but also invariant along the magnetic field . thus , the liquid again oscillates as solid columns , but in this case transversely to the field which has no effect on such a flow . this is because the e.m.f induced by the flow , which is invariant along the magnetic field , is irrotational , i.e. , @xmath152 and , thus , unable to drive current circulation in a closed liquid volume . note that for the axisymmetric modes @xmath153 the potential ( [ sol : phi0-p ] ) and the associated velocity ( [ sol : v0-e ] ) take an indeterminate form . namely , for @xmath154 boundary condition ( [ eq : r0-e ] ) , which in this is case straightforwardly implies a zero frequency , is satisfied by an arbitrary potential distribution independent of the radius perturbation . as seen from ( [ sol : phi0-p ] ) , a non - zero axisymmetric potential is associated with a purely azimuthal velocity . consequently , this mode is irrelevant and can subsequently be neglected because it represents an internal flow perturbation which is just compatible but not coupled with axisymmetric shape deformations similarly to the fast modes discussed at the end of @xmath6[sec : invisc ] . moreover , this is consistent with ( [ sol : bt0-e ] ) according to which axisymmetric transversal modes are static in the leading - order approximation that implies a zero velocity and , consequently , a zero associated potential . expression ( [ sol : v0-e ] ) implies that the velocity streamlines coincide with the isolines of @xmath155 which , thus , represents a stream function for the flow oscillations . figure [ fig : even ] shows the shapes and streamlines of the associated liquid flow in the horizontal mid - plane for the first four transversal oscillation modes . note that the first and the third mode with the indices @xmath156 and @xmath157 are both naturally invariant in the direction of the magnetic field and , thus , effectively non - magnetic . the second mode with @xmath158 corresponds to the drop oscillating in such a way that horizontal cross - sections remain circular in the small - amplitude limit under consideration while the whole shape deforms because of vertical offset of the cross - sections . shapes and the associated liquid flows in the horizontal mid - plane @xmath159 perpendicular to the magnetic field for the first four transversal oscillation modes defined by indices @xmath156 ( a ) , @xmath160 ( b ) , ( 3,3 ) ( c ) and @xmath161 ( d).,title="fig:",scaledwidth=25.0%](-110,0)(a)shapes and the associated liquid flows in the horizontal mid - plane @xmath159 perpendicular to the magnetic field for the first four transversal oscillation modes defined by indices @xmath156 ( a ) , @xmath160 ( b ) , ( 3,3 ) ( c ) and @xmath161 ( d).,title="fig:",scaledwidth=25.0%](-110,0)(b)shapes and the associated liquid flows in the horizontal mid - plane @xmath159 perpendicular to the magnetic field for the first four transversal oscillation modes defined by indices @xmath156 ( a ) , @xmath160 ( b ) , ( 3,3 ) ( c ) and @xmath161 ( d).,title="fig:",scaledwidth=25.0%](-110,0)(c)shapes and the associated liquid flows in the horizontal mid - plane @xmath159 perpendicular to the magnetic field for the first four transversal oscillation modes defined by indices @xmath156 ( a ) , @xmath160 ( b ) , ( 3,3 ) ( c ) and @xmath161 ( d).,title="fig:",scaledwidth=25.0%](-110,0)(d ) in order to determine the magnetic damping rates for longitudinal modes , we have to consider the first - order solution governed by@xmath162 which yields@xmath163 then ( [ eq : phih - cnd ] ) and ( [ eq : r0-o ] ) applied consecutively result in @xmath164 which combined with ( [ eq : rh - cnd ] ) , ( [ eq : ph - cnd ] ) and ( [ eq : r0-o ] ) yields @xmath165z^{-1}\hat{r}_{0}^{o}+\beta_{0}^{o}(2\beta_{1}^{o}-m^{2}{\beta_{0}^{o}}^{2})\hat{r}_{0}^{o}.\ ] ] after some algebra , we obtain @xmath166z^{-1}\right\|_{r=1}\equiv l_{z}+2+m^{2},$ ] and , consequently , @xmath167\hat{r}_{0}^{o}.\label{egv : r1-o}\ ] ] the l.h.s . operator above is the same as that in ( [ egv : r0-o ] ) which has @xmath118 as its eigensolution with a zero eigenvalue . owing to ( [ eq : legendre ] ) and ( [ sol : r0-o ] ) , @xmath118 is an eigensolution of the r.h.s operator of ( [ egv : r1-o ] ) , too . thus , for ( [ egv : r1-o ] ) to be solvable , its r.h.s has to be free of the terms proportional to @xmath168 that yields @xmath169 note that conversely to the frequency for longitudinal oscillation modes ( [ sol : bt0-o ] ) , the magnetic damping rate above is not degenerate and varies with @xmath100 similarly to the oscillation frequency considered above , boundary conditions ( [ eq : ph - cnd ] ) and ( [ eq : rh - cnd ] ) applied to the first - order solution ( [ eq : p - phi-1e ] ) result in @xmath170\hat{r}_{0}^{e},\label{eq : r1-e}\\ ( l_{z}+2)\hat{r}_{1}^{e } & = & -\hat{p}_{1}^{e}.\label{eq : p1-e}\end{aligned}\ ] ] to solve this first - order problem , we again need a second - order solution governed by@xmath171 which , by taking into account ( [ eq : p - phi-1e ] ) , yields @xmath172\hat{\varphi}_{0}^{+}-\frac{1}{2}\beta_{1}^{e}z^{2}l_{r}\hat{\varphi}_{0}^{+},\label{eq : p - phi-2e}\ ] ] then ( [ eq : ph - cnd ] ) results in@xmath173\hat{r}_{0}^{e}.\label{eq : phi2-e}\ ] ] substituting @xmath174 and @xmath175 from ( [ eq : r1-e ] ) and ( [ eq : phi2-e ] ) into ( [ eq : p1-e ] ) and using @xmath176 after some algebra we obtain an equation for @xmath177 which is the same as ( [ egv : r0-e ] ) for @xmath178 except for the r.h.s . that now reads as @xmath179\hat{r}_{0}^{e}\ ] ] by the same arguments as for ( [ egv : r1-o ] ) , the solvability condition applied to the expression above results in@xmath180 which again coincides with the corresponding result of @xcite . there are three effects due to viscosity in this problem . firstly , viscosity appears in the normal stress balance condition ( [ eq : nblnc - nd ] ) as a @xmath181 correction to the inviscid solution obtained above . secondly , viscosity also appears as a small parameter @xmath182 in ( [ eq : phip ] ) which again implies the same order correction when the leading - order inviscid solution is substituted into this term . thirdly , viscosity enters the problem implicitly through the free - slip boundary condition ( [ eq : tblnc ] ) which was ignored by the inviscid solution but needs to be satisfied when viscosity is taken into account . to satisfy this condition , the leading - order solution needs to be corrected by the viscous term in ( [ eq : phip ] ) , where @xmath182 appears as a small parameter at the higher - order derivative . for this small viscous term to become comparable with the dominating magnetic term at the surface , the expected correction has to vary over the characteristic length scale @xmath183 which is defined by the hartmann number @xmath184 . moreover , for the viscous correction of the tangential velocity @xmath185 in the hartmann layer to compensate for a @xmath186 tangential stress due to the leading - order inviscid solution , @xmath187 is required . then the incompressibility constraint implies an associated normal velocity component of an order in @xmath188 smaller than @xmath189 i.e. , @xmath190 this normal velocity correction is subsequently negligible . but this not the case for the tangential velocity correction @xmath189 which according to ( [ eq : p ] ) is expected to produce a pressure correction @xmath191 the last is comparable with the normal viscous stress produced by the leading - order inviscid flow . taking into account the estimates above and @xmath192 which follows from ( [ eq : phi ] ) , we search for a viscous correction as @xmath193 where the terms with the tilde account for a hartmann layer solution localised at the surface . we start with the core region , where the additive boundary layer corrections are supposed to vanish . the first - order viscous corrections for the pressure and potential @xmath194 are obtained similarly to the leading - order inviscid solution ( [ eq : p0f0 ] ) . now , instead of the kinematic and electric boundary conditions ( [ eq : r1-cnd ] ) and ( [ eq : phi - cnd ] ) derived in the inviscid approximation , we have to use the original ones ( [ eq : kinc ] ) and ( [ eq : jr ] ) containing the velocity , which again follows from the navier - stokes equation ( [ eq : ns - nd ] ) including the viscous term @xmath195 for the longitudinal modes , described by the odd solutions , ( [ eq : ns - nd ] ) yields @xmath196 where @xmath197 and @xmath198 are the velocity components parallel and perpendicular , respectively , to the field direction @xmath199 and @xmath200 is a spectral counterpart of the nabla operator for the azimuthal mode @xmath100 since , as shown above , both the potential and velocity perturbations in the hartmann layer are higher - order small quantities and , thus , negligible with respect to the core perturbation , the electric boundary condition ( [ eq : jr ] ) can be applied at @xmath87 directly to the first - order core solution as @xmath201 taking into account ( [ eq : u01-o ] ) , this yields @xmath202 and , hence , @xmath203 consequently , the first - order velocity perturbation in the core for the odd modes is again purely longitudinal . then , the kinematic constraint ( [ eq : kinc ] ) for the leading- and first - order terms takes , respectively , the form@xmath204 these expressions combined with ( [ eq : w01-o ] ) result in@xmath205 which defines the first - order core pressure perturbation at @xmath206 in addition , we need also the hartmann layer pressure correction which according to the estimates above is of the same order of magnitude as the core one . to resolve the hartmann layer , we introduce a stretched coordinate @xmath207 ( @xcite ) , where @xmath208 is the characteristic hartmann layer thickness . in the hartmann layer variables , ( [ eq : phip ] ) takes the form@xmath209 for @xmath210 , the inertial term @xmath211 is negligible in ( [ eq : pfv - bnd ] ) with respect to the magnetic one @xmath212 except for @xmath213 first , ignoring this term , which , as shown below , gives a next - order small correction , the solution of ( [ eq : pfv - bnd ] ) vanishing outside the hartmann layer can be written as @xmath214 where the index @xmath215 denotes the surface distribution of the corresponding quantity . then the free - slip boundary condition ( [ eq : tblnc ] ) results in@xmath216 for the longitudinal modes , defined by the odd solutions , the leading - order inviscid velocity is purely axial @xmath217 substituting this into ( [ eq : vs - theta ] ) and taking into account that the radial pressure distribution at the surface is related to the radius perturbation by ( [ eq : r0-o ] ) , we obtain@xmath218 pressure is related to the velocity by ( [ eq : p ] ) , which in the dimensionless form reads as @xmath219 in the hartmann layer variables , this equation takes the form @xmath220 substituting the general solutions for pressure and velocity given by ( [ eq : pfv - s ] ) into ( [ eq : p - bnd ] ) and using ( [ eq : vs - theta - o ] ) , we find@xmath221 substituting the normal component of viscous stress @xmath222 together with the core and boundary layer pressure contributions defined by ( [ eq : p01-o ] ) and ( [ eq : ps - o ] ) into the normal stress balance condition ( [ eq : nblnc - nd ] ) , we finally obtain@xmath223.\label{eq : r01-o}\ ] ] the sought for viscous damping rate is obtained in the usual way by applying the solvability condition to ( [ eq : r01-o ] ) that after some algebra results in@xmath224\frac{p_{l}^{m}(z)}{z}\ , dz=-(2l+1)\left[\frac{1}{2}(l(l+1)-m^{2})-1-i_{l}^{m}\right],\nonumber \end{aligned}\ ] ] where @xmath225 can be calculated from the above recurrence relation starting with @xmath226 and taking into account that @xmath227 for @xmath228 for the modes with @xmath229 , we have @xmath230 which is the half of the corresponding viscous damping rate without the magnetic field ( @xcite ) . although the viscous damping rate increases for smaller @xmath84 as seen from the numerical values of @xmath231 for the first 7 longitudinal oscillation modes calculated by the mathematica ( @xcite ) and shown in table [ tab : bet01-o ] , it remains below its non - magnetic counterpart up to @xmath232 modes . note that the r.h.s of ( [ eq : r01-o ] ) has a simple pole @xmath233 singularity at @xmath234 which is due to the neglected inertial term in ( [ eq : pfv - bnd ] ) . as discussed above , this term becomes relevant for @xmath235 where it cuts off the singularity at @xmath236 this cut - off integrated in ( [ sol : bet01-o ] ) over @xmath235 where @xmath237 for the odd modes , results in the damping rate correction @xmath238 which is a higher - order small quantity . .[tab : bet01-o]the viscous damping rates @xmath231 for the first 6 longitudinal oscillation modes . [ cols="^,^,^,^,^,^,^,^",options="header " , ] the approach above is not directly applicable to the axisymmetric transversal modes which , as discussed at the end of @xmath6[sub : t - freq ] , are stationary in the leading - order inviscid approximation . for these overdamped modes , a flow with the velocity @xmath239 relative to the leading - order radius perturbation appears only in the first - order approximation , which according to ( [ eq : nrg ] ) produces the same order ohmic dissipation . in this case , dissipation takes place on the account of the surface energy reduction , while that of the kinetic energy is negligible because it is by @xmath240 smaller than the former . the contribution of the viscous dissipation in ( [ eq : nrg ] ) is @xmath241 which for a low viscosity and a high magnetic field is much smaller than the ohmic dissipation @xmath242 and , thus negligible with respect to the latter . in the present study , we have considered small - amplitude oscillations of a conducting liquid drop in a uniform dc magnetic field . viscosity was assumed to be small but the magnetic field strong . combining the regular and matched asymptotic expansion techniques we obtained a relatively simple solution to the associated eigenvalue problem . firstly , we showed that the eigenmodes of shape oscillations are not affected by strong magnetic field . namely , they remain the spherical harmonics as in the non - magnetic case . strong magnetic field , however , constrains the liquid flow associated with the oscillations and , thus , reduces the oscillations frequency by increasing apparent inertia of the liquid . in such a field , liquid oscillates in a two - dimensional ( 2d ) way as solid columns aligned with the field . two types of oscillations are possible : longitudinal and transversal to the field . such oscillations are weakly damped by strong magnetic field the stronger the field , the weaker the damping , except for the axisymmetric transversal and 2d modes . the former are magnetically overdamped because the incompressibility constraint does not permit an axially uniform radial flow . the latter , which are transversal modes defined by the spherical harmonics with equal degree and order , @xmath5 , are not affected by the magnetic field because these modes are naturally invariant along the field . in a uniform magnetic field , no electric current is induced and , thus , no electromagnetic force acts on such a 2d transversal flow because the associated e.m.f . is irrotational . because the magnetic damping for all other modes decreases inversely with the square of the field strength , the viscous damping may become important in a sufficiently strong magnetic field . consequently , the relaxation of axisymmetric transversal modes , whose viscous damping is negligible relative to the magnetic one , can be used to determine the electrical conductivity , while the damping of @xmath5 modes can be used to determine the viscosity . the damping of all other modes is affected by both the viscous and ohmic dissipations . although the latter reduces inversely with the square of the field strength while the former stays constant , an extremely strong magnetic field may be required for the viscous dissipation to be become dominant . as an example , let us consider a drop of nickel of @xmath243 in diameter @xmath244 which , at the melting point @xmath245 has the surface tension @xmath246 density @xmath247 the dynamic viscosity @xmath248 and the electrical conductivity @xmath249 ( @xcite ) . the capillary time scale and frequency of non - magnetic fundamental mode @xmath250 for such a drop are @xmath251 and @xmath252 respectively . the viscous damping time without the magnetic field ( @xcite ) is @xmath253 , where @xmath254 is the viscous time scale . note that weak - viscosity approximation is applicable in this case because @xmath255 is small . in the magnetic field of @xmath256 for which @xmath257 the oscillation frequency of longitudinal fundamental mode @xmath131 drops according to equation ( [ sol : bt0-o ] ) to @xmath258 the corresponding viscous damping time increases by the factor of two to @xmath259 where @xmath260 according to table [ tab : bet01-o ] . the magnetic damping time of this mode , for which ( [ eq : bt1-o ] ) yields @xmath261 is @xmath262 according to this formula , for the magnetic damping time to exceed the viscous one , a magnetic field of @xmath263 is necessary . the relaxation time for the axisymmetric fundamental mode @xmath264 which is magnetically over - damped , is @xmath265 where @xmath266 follows from ( [ eq : bt1-e ] ) . the magnetic field affects neither the frequency nor the damping rate of @xmath156 transversal oscillation mode , which is naturally invariant along the field . for the same reason , there is no magnetic damping of this mode either . the first oscillatory transversal mode is @xmath158 whose frequency drops according to ( [ sol : bt0-e ] ) from @xmath267 without the magnetic field to @xmath268 in a strong magnetic field . the magnetic damping time for this mode in a @xmath269 magnetic field is @xmath270 where @xmath271 follows from ( [ eq : bt1-e ] ) . the viscous damping time for this mode is @xmath272 where @xmath273 follows from table [ tab : bet01-e ] . the viscous damping is small relative to the magnetic one for this mode , and a magnetic field of about @xmath274 would be necessary for the magnetic damping time to become as long as the viscous one . in conclusion , this theoretical model provides a basis for the development of new measurement method of surface tension , viscosity and electrical conductivity of liquid metals using oscillating drop technique in a strong superimposed dc magnetic field . fukuyama , h. , takahashi , k. , sakashita , s. , kobatake , h. , tsukada , t. & awaji , s. 2009 noncontact modulated laser calorimetry for liquid austenitic stainless steel in dc magnetic field . _ isij int . _ * 49 * , 14361442 . rhim , w .- k . , ohsaka , k. , paradis , p .- f . & spjut , r. e. 1999 noncontact technique for measuring surface tension and viscosity of molten materials using high temperature electrostatic levitation . instrum . _ * 70 * , 27962801 . tsukada , t. , sugioka , k. , tsutsumino , t. , fukuyama , h. , kobatake , h. 2009 effect of static magnetic field on a thermal conductivity measurement of a molten droplet using an electromagnetic levitation technique . j. heat mass tran . _ * 52 * , 51525157 . yasuda , h. , ohnaka , i. , ninomiya , y. , ishii , r. , fujita , s. , kishio , k. , fujita , s. & kishio , k. 2004 levitation of metallic melt by using the simultaneous imposition of the alternating and the static magnetic fields . _ j. cryst . growth _ * 260 * , 475485 . yasuda , h. , ohnaka , i. , ishii , r. , fujita , s. , & tamura , y. 2005 investigation of the melt flow on solidified structure by a levitation technique using alternative and static magnetic fields . _ isij int . _ * 45 * , 991996 . yasuda , h. 2007 applications of high magnetic fields in materials processing , _ in : molokov , s. , moreau , r. , moffatt , h.k . ( eds . ) , magnetohydrodynamics historical evolution and trends . _ springer , 329344 .	we analyse small - amplitude oscillations of a weakly viscous electrically conducting liquid drop in a strong uniform dc magnetic field . an asymptotic solution is obtained showing that the magnetic field does not affect the shape eigenmodes , which remain the spherical harmonics as in the non - magnetic case . strong magnetic field , however , constrains the liquid flow associated with the oscillations and , thus , reduces the oscillation frequencies by increasing effective inertia of the liquid . in such a field , liquid oscillates in a two - dimensional ( 2d ) way as solid columns aligned with the field . two types of oscillations are possible : longitudinal and transversal to the field . such oscillations are weakly damped by a strong magnetic field the stronger the field , the weaker the damping , except for the axisymmetric transversal and inherently 2d modes . the former are overdamped because of being incompatible with the incompressibility constraint , whereas the latter are not affected at all because of being naturally invariant along the field . since the magnetic damping for all other modes decreases inversely with the square of the field strength , viscous damping may become important in a sufficiently strong magnetic field . the viscous damping is found analytically by a simple energy dissipation approach which is shown for the longitudinal modes to be equivalent to a much more complicated eigenvalue perturbation technique . this study provides a theoretical basis for the development of new measurement methods of surface tension , viscosity and the electrical conductivity of liquid metals using the oscillating drop technique in a strong superimposed dc magnetic field .
we have presented in detail the calculation of @xmath14 radiative corrections to the pair production of charginos and neutralinos within the minimal supersymmetric standard model . we discussed a possible separation of weak and qed corrections . although only the sum of virtual and real bremsstrahlung corrections have physical meaning , a separation of these two contributions has its advantages . the qed corrections are treated numerically by integration over the phase space with an additional photon in the final state . quite often a cancellation between qed and weak corrections occurs leading to a small total correction . furthermore , we used the structure function formalism to include higher order qed corrections and to split the qed corrections into a universal and a non - universal part . the non - universal qed corrections contain the integration over the three particle final state and are in the case of the total cross - section often negligible . the numerical analysis is based on the sps1a benchmark scenario , proposed within the spa project . the @xmath2 parameters of the spa convention are translated into a set of on - shell parameters , which serve as input for our on - shell renormalization scheme . such translation tools make it possible to compare cross sections , masses , etc . calculated in different renormalization schemes . the presented numerics shows results for the total cross - section , left - right , and forward - backward asymmetries . in all three cases we have sizeable radiative corrections around the 10% and in some cases even higher . we thank for useful correspondence with w. porod and t. fritzsche . the authors acknowledge support from eu under the hprn - ct-2000 - 00149 network programme . the work was also supported by the `` fonds zur frderung der wissenschaftlichen forschung '' of austria , project no . p13139-phy . 99 j. a. aguilar - saavedra _ et al . _ [ ecfa / desy lc physics working group ] , [ arxiv : hep - ph/0106315 ] ; k. abe _ et al . _ [ acfa linear collider working group ] , [ arxiv : hep - ph/0109166 ] ; t. abe _ et al . _ [ american linear collider working group ] , in _ proc . of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed . n. graf , [ arxiv : hep - ex/0106056 ] . s. y. choi , a. djouadi , m. guchait , j. kalinowski , h. s. song and p. m. zerwas , eur . j. c * 14 * ( 2000 ) 535 [ arxiv : hep - ph/0002033 ] ; s. y. choi , m. guchait , j. kalinowski and p. m. zerwas , phys . b * 479 * ( 2000 ) 235 [ arxiv : hep - ph/0001175 ] . s. y. choi , j. kalinowski , g. moortgat - pick and p. m. zerwas , eur . j. c * 22 * ( 2001 ) 563 [ addendum - ibid . c * 23 * ( 2002 ) 769 ] [ arxiv : hep - ph/0108117 ] . j. kblbeck , m. bhm , a. denner , comput . * 60 * ( 1990 ) 165 ; t. hahn , comput . * 140 * ( 2001 ) 418 ; t. hahn , c. schappacher , comput . * 143 * ( 2002 ) 54 ; t. hahn , m. perez - victoria , comput . phys . commun . * 118 * ( 1999 ) 153 . h. eberl , m. kincel , w. majerotto and y. yamada , nucl . b * 625 * ( 2002 ) 372 [ arxiv : hep - ph/0111303 ] ; c. weber , h. eberl , w. majerotto phys . lett . b * 572 * ( 2003 ) 56 [ arxiv : hep - ph/0305250 ] ; c. weber , h. eberl , w. majerotto phys d * 68 * ( 2003 ) 093011 [ arxiv : hep - ph/0308146 ] . h. eberl , m. kincel , w. majerotto , y. yamada , phys . d * 64 * ( 2001 ) 115013 [ arxiv : hep - ph/0104109 ] ; w. ller , h. eberl , w. majerotto , c. weber , eur . j. c * 29 * ( 2003 ) 563 [ arxiv : hep - ph/0304006 ] . e. a. kuraev and v. s. fadin , sov . j. nucl . * 41 * ( 1985 ) 466 [ yad . fiz . * 41 * ( 1985 ) 733 ] ; g. altarelli and g. martinelli , in `` physics at lep '' , eds . j. ellis and r. peccei , ( cern 86 - 02 , geneva , 1986 ) , vol . 1 , p.47 ; f. a. berends , w. l. van neerven and g. j. h. burgers , nucl . b * 297 * ( 1988 ) 429 [ erratum - ibid . b * 304 * ( 1988 ) 921 ] ; w. beenakker , f.a . berends and w.l . van neerven , in `` radiative corrections for @xmath0 collisions '' , ed . khn ( 1989 ) p.3 .	we present a high precision calculation of chargino and neutralino pair production at @xmath0 colliders . within the minimal supersymmetric standard model , the full one - loop and higher order qed corrections are included . special care has been taken in the definition of the lagrangian input parameters . furthermore , the proper inclusion of qed corrections and the separation of weak and qed corrections are discussed . we show numerical results for total cross sections , as well as forward - backward and left - right asymmetries for the sps1a scenario as proposed in the spa project . the complete corrections are about 10% and in some cases even larger , in particular for @xmath1 production with sizeable higgsino components . these corrections have to be taken into account in a high precision analysis . hephy - pub 806/05 + hep - ph/0504109 * precise predictions for chargino and neutralino pair production in @xmath0 annihilation * w. ller , h. eberl , w. majerotto [ cols= " < " , ] for all other parameters that are free of renormalization conditions , the or on - shell values can be used . the difference is of higher order for the current processes . + since we use an on - shell renormalization , the appropriate tree - level for the one - loop calculation is given in terms of on - shell parameters . on the other hand , our original input are the @xmath2 parameters of the sps1a scenario . we therefore show the corrections relative to the tree - level in the spa convention , i.e. calculated in terms of on - shell masses and @xmath2 values for all couplings . using this tree - level definition , the relative corrections are the same compared with those calculated in other renormalization schemes , up to terms of higher order . for this purpose the used @xmath2 value at @xmath3=1 tev of the fine structure constant is @xmath4 . in the presented numerics , we use for the charge renormalization the @xmath5 scheme for neutralino production and the @xmath6 scheme for chargino production , as discussed in section [ gauge sector ] . by calculating the same cross section in both schemes , we find good agreement within a few per - mill in the final results . the figs . ( [ fig : cha_tot1 ] ) and ( [ fig : neu_tot12 ] ) show the total cross sections for chargino and neutralino pair production at the tree - level in the spa convention , and with weak and full corrections . additionally , fig . ( [ fig : cha_tot1 ] ) shows the complete corrections to the improved tree - level , where isr is already included . since the soft - photon pole is already absorbed into the tree - level , we have moderate corrections even near the threshold . comparing the various corrections to the total tree - level cross section for the different production channels figs . ( [ fig : cha_corr12 ] , [ fig : cha_corr34 ] , [ fig : neu_corr12 ] ) some common characteristics can be recognized . near the threshold the negative soft photon contributions dominate . far away from the threshold , the positive universal qed corrections partially cancel the large and negative weak contributions . the almost constant non - universal qed corrections are small and in comparison with other corrections often negligible . due to the marginal non - universal qed corrections , the differences between the two proposed ways to highlight genuine weak " corrections are quite small . however , this does not have to be true any longer if cuts on the phase space are applied or distributions in particle energies or scattering angles are discussed . moreover , we present results for the forward - backward and left - right asymmetry @xmath7 and @xmath8 of chargino production in the figs . ( [ fig : cha_afb1 ] , [ fig : cha_alr1 ] ) . these are defined by @xmath9 where @xmath10 with @xmath11 is the angle between the incoming electron and the outgoing @xmath12 in the cms frame . @xmath13 denotes the total cross section for left / right - handed electrons and unpolarized positrons . both asymmetries obtain sizeable corrections for all three production channels . the kinks in the lines can be traced back to so - called normal and anomalous thresholds .
directed planar graphs with weighted edges have been widely used in the study of totally nonnegative matrices ( @xcite ) . ( reviews of the area can be found in @xcite . ) in particular , a special kind of such graphs is a convenient tool for visualizing in the @xmath0 case lusztig type parametrizations of double bruhat cells @xcite . each parametrization of this kind is obtained via a factorization of an element of a cell into a product of elementary factors . the standard poisson - lie structure on reductive lie groups is induced by a very simple poisson bracket on factorization parameters . recently , postnikov used weighted directed planar graphs to parametrize cells in grassmannians @xcite . the main goal of this paper is to investigate poisson properties of postnikov s parametrization . using a graphical interpretation of the poisson - lie property inspired by the lusztig parametrization , we introduce a natural family of poisson brackets on the space of edge weights . these brackets induce a two - parameter family of poisson brackets on the grassmannian . every poisson bracket in this family is compatible ( in the sense of @xcite ) with the cluster algebra on the grassmannian described in @xcite and , on the other hand , endows the grassmannian with a structure of a poisson homogeneous space with respect to the natural action of the general linear group equipped with an r - matrix poisson - lie structure . this paper is the first in the series devoted to the study of poisson properties of weighted directed graphs on surfaces . the second paper @xcite generalizes results of the current one to graphs in an annulus . in this case , the analogue of postnikov s construction leads to a map into the space of loops in the grassmannian . natural poisson brackets on edge weights in this case are intimately connected to trigonometric r - matrix brackets on matrix - valued rational functions . the third paper @xcite utilizes particular graphs in an annulus to introduce a cluster algebra structure related to the coordinate ring of the space of normalized rational functions in one variable . this space is birationally equivalent , via the moser map @xcite , to any minimal irreducible coadjoint orbit of the group of upper triangular matrices associated with a coxeter element of the permutation group . in this case , the poisson bracket compatible with the cluster algebra structure coincides with the quadratic poisson bracket studied in @xcite in the context of toda flows on minimal orbits . we show that cluster transformations serve as bcklund - darboux transformations between different minimal toda flows . the fourth paper @xcite solves , in the case of graphs in an annulus with one source and one sink , the inverse problem of restoring the weights from the image of the generalized postnikov map . in the case of arbitrary planar graphs in a disk , this problem was completely solved by postnikov @xcite who proved that for a fixed minimal graph , the space of weights modulo gauge action is birational to its image and described all minimal graphs . to the contrary , already for simplest graphs in an annulus , the corresponding map can only be shown to be finite . in @xcite , the inverse problem for the totally nonnegative matrices is solved for a particular type of minimal graphs . in @xcite we describe all minimal graphs with one source and one sink . the paper is organized as follows . in section [ ppn ] , we introduce a notion of a perfect network in a disk and associate with every such network a _ matrix of boundary measurements_. each boundary measurement is shown to be a rational function in edge weights and admits a subtraction - free rational expression , see proposition [ sfree ] . in section [ psmat ] , we characterize all _ universal _ poisson brackets on the space of edge weights of a given network that respect the natural operation of concatenation of networks , see proposition [ 6param ] . furthermore , we establish that the family of universal brackets induces a linear two - parameter family of poisson brackets on boundary measurement matrices , see theorem [ psm ] . this family depends on a mutual location of sources and sinks , but not on the network itself . we provide an explicit description of this family in theorem [ psme ] . in section [ psongr ] , we start by showing that if the sources and the sinks are not intermixed along the boundary of the disk , one can recover a 2-parametric family of r - matrices and the corresponding r - matrix brackets on @xmath0 , see theorem [ psgl ] . next , the boundary measurement map defined by a network with @xmath1 sources and @xmath2 sinks is extended to the grassmannian boundary measurement map into the grassmannian @xmath3 . the poisson family on boundary measurement matrices allows us to equip the grassmannian with a two - parameter family of poisson brackets @xmath4 in such a way that for any choice of a universal poisson bracket on edge weights there is a unique member of @xmath4 that makes the grassmannian boundary measurement map poisson , see theorem [ psgr ] . this latter family depends only on the number of sources and sinks and does not depend on their mutual location . finally , we give an interpretation of the natural @xmath0 action on @xmath3 in terms of networks and establish that every member of the poisson family @xmath4 on @xmath3 makes @xmath3 into a poisson homogeneous space of @xmath0 equipped with the above described r - matrix bracket , see theorem [ poihomo ] . in section [ compclust ] , we review the construction of the cluster algebra structure on an open cell in the grassmannian given in @xcite . we then introduce face weights and use them to show that every member of the poisson family @xmath4 is compatible with this cluster algebra structure , see theorem [ thm : poisscomp ] . let @xmath5 be a directed planar graph with no loops and parallel edges drawn inside a disk ( and considered up to an isotopy ) with the vertex set @xmath6 and the edge set @xmath7 . exactly @xmath8 of its vertices are located on the boundary circle of the disk . they are labelled counterclockwise @xmath9 and called _ boundary vertices _ ; occasionally we will write @xmath10 for @xmath11 and @xmath12 for @xmath13 . each boundary vertex is marked as a source or a sink . a _ source _ is a vertex with exactly one outcoming edge and no incoming edges . _ sinks _ are defined in the same way , with the direction of the single edge reversed . the number of sources is denoted by @xmath1 , and the corresponding set of indices , by @xmath14 $ ] ; the set of the remaining @xmath15 indices is denoted by @xmath16 . all the internal vertices of @xmath17 have degree @xmath18 and are of two types : either they have exactly one incoming edge , or exactly one outcoming edge . the vertices of the first type are called ( and shown on figures ) _ white _ , those of the second type , _ black_. let @xmath19 be independent variables . a _ perfect planar network _ @xmath20 is obtained from @xmath17 as above by assigning a weight @xmath21 to each edge @xmath22 . in what follows we occasionally write `` network '' instead of `` perfect planar network '' . each network defines a rational map @xmath23 ; the _ space of edge weights _ @xmath24 is defined as the intersection of the image of @xmath25 under @xmath26 with @xmath27 . in other words , a point in @xmath24 is a graph @xmath17 as above with edges weighted by nonzero reals obtained by specializing the variables @xmath19 in the expressions for @xmath28 to nonzero values . an example of a perfect planar network is shown in fig . [ fig : pergraph ] . it has two sources : @xmath13 and @xmath29 , and two sinks @xmath30 and @xmath31 . each edge @xmath32 is labelled by its weight . the weights depend on four independent variables @xmath33 and are given by @xmath34 the space of edge weights is the 4-dimensional subvariety in @xmath35 given by equations @xmath36 , @xmath37 , @xmath38 , @xmath39 and condition @xmath40 . a _ path _ @xmath41 in @xmath42 is an alternating sequence @xmath43 of vertices and edges such that @xmath44 for any @xmath45 $ ] . sometimes we omit the names of the vertices and write @xmath46 . a path is called a _ cycle _ if @xmath47 and a _ simple cycle _ if additionally @xmath48 for any other pair @xmath49 . to define the weights of the paths we need the following construction . consider a closed oriented polygonal plane curve @xmath50 . let @xmath51 and @xmath52 be two consequent oriented segments of @xmath50 , and let @xmath53 be their common vertex . we assume for simplicity that for any such pair @xmath54 , the cone spanned by @xmath51 and @xmath52 is not a line ; in other words , if @xmath51 and @xmath52 are collinear , then they have the same direction . observe that since @xmath50 is not necessary simple , there might be other edges of @xmath50 incident to @xmath53 ( see figure [ fig : concord ] below ) . let @xmath55 be an arbitrary oriented line . define @xmath56 in the following way : @xmath57 if the directing vector of @xmath55 belongs to the interior of the cone spanned by @xmath51 and @xmath52 and @xmath58 otherwise ( see figure [ fig : concord ] for examples ) . define @xmath59 as the sum of @xmath60 over all pairs of consequent segments in @xmath50 . it follows immediately from theorem 1 in @xcite that @xmath59 does not depend on @xmath55 , provided @xmath55 is not collinear to any of the segments in @xmath50 . the common value of @xmath59 for different choices of @xmath55 is denoted by @xmath61 and called the _ concordance number _ of @xmath50 . in fact , @xmath61 equals @xmath62 the rotation number of @xmath50 ; the definition of the latter is similar , but more complicated . ; @xmath63 ; @xmath64,height=170 ] in what follows we assume without loss of generality that @xmath42 is drawn in such a way that all its edges are straight line segments and all internal vertices belong to the interior of the convex hull of the boundary vertices . given a path @xmath41 between a source @xmath65 and a sink @xmath66 , we define a closed polygonal curve @xmath67 by adding to @xmath41 the path between @xmath66 and @xmath65 that goes counterclockwise along the boundary of the convex hull of all the boundary vertices of @xmath42 . finally the weight of @xmath41 is defined as @xmath68 the weight of an arbitrary cycle in @xmath42 is defined in the same way via the concordance number of the cycle . if edges @xmath32 and @xmath69 in @xmath41 coincide and @xmath70 , the path @xmath41 can be decomposed into the path @xmath71 and the cycle @xmath72 . clearly , @xmath73 , and hence @xmath74 consider the path @xmath75 in figure [ fig : pergraph ] . choose @xmath55 as shown in the figure ; @xmath55 is neither collinear with the edges of @xmath41 nor with the relevant edges of the convex hull of boundary vertices ( shown by dotted lines ) . clearly , @xmath58 for all pairs of consecutive edges of @xmath67 except for the pairs @xmath76 and @xmath77 , where @xmath78 is the additional edge joining @xmath30 and @xmath31 . so , @xmath79 , and hence @xmath80 . the same result can be obtained by decomposing @xmath41 into the path @xmath81 and the cycle @xmath82 . [ dualdir]instead of closed polygonal curves @xmath67 , one can use curves @xmath83 obtained by adding to @xmath41 the path between @xmath66 and @xmath65 that goes clockwise along the boundary of the convex hull of all the boundary vertices of @xmath42 . it is a simple exercise to prove that the concordance numbers of @xmath67 and @xmath83 coincide . therefore , the weight of a path can be defined also as @xmath84 given a perfect planar network as above , a source @xmath65 , @xmath85 , and a sink @xmath66 , @xmath86 , we define the _ boundary measurement _ @xmath87 as the sum of the weights of all paths starting at @xmath65 and ending at @xmath66 . clearly , the boundary measurement thus defined is a formal infinite series in variables @xmath28 , @xmath22 . however , this series possesses certain nice propetsies . recall that a formal power series @xmath88 $ ] is called a rational function if there exist polynomials @xmath89 $ ] such that @xmath90 in @xmath91 $ ] . in this case we write @xmath92 . for example , @xmath93 in @xmath94 $ ] . besides , we say that @xmath95 admits a _ subtraction - free _ rational expression if it can be written as a ratio of two polynomials with nonnegative coefficients . for example , @xmath96 admits a subtraction - free rational expression since it can be written as @xmath97 . the following result was proved in @xcite[lemma 4.3 ] and further generalized in @xcite , but we will present an alternative proof to illustrate the method that will be used in other proofs below . [ sfree ] let @xmath42 be a perfect planar network in a disk , then each boundary measurement in @xmath42 is a rational function in the weights @xmath28 admitting a subtraction - free rational expression . we prove the claim by induction on the number of internal vertices . the base of induction is the case when there are no internal vertices at all , and hence each edge connects a source and a sink ; in this case the statement of the proposition holds trivially . assume that @xmath42 has @xmath98 internal vertices . consider a specific boundary measurement @xmath87 . the claim concerning @xmath87 is trivial if @xmath66 is the neighbor of @xmath65 . in the remaining cases @xmath65 is connected by an edge @xmath99 to its only neighbor in @xmath17 , which is either white or black . assume first that the neighbor of @xmath65 is a white vertex @xmath53 . create a new network @xmath100 by deleting @xmath65 and the edge @xmath101 from @xmath17 , splitting @xmath53 into @xmath102 sources @xmath103 ( so that @xmath104 in the counterclockwise order ) and replacing the edges @xmath105 and @xmath106 by @xmath107 and @xmath108 , respectively , both of weight 1 ( see figure [ fig : whitefork ] ) . clearly , to any path @xmath41 from @xmath65 to @xmath66 corresponds either a path @xmath109 from @xmath110 to @xmath66 or a path @xmath111 from @xmath112 to @xmath66 . moreover , @xmath113 . therefore @xmath114 where @xmath115 means that the measurement is taken in @xmath100 . observe that the number of internal vertices in @xmath100 is @xmath116 , hence the claim follows from the above relation by induction . assume now that the neighbor of @xmath65 is a black vertex @xmath117 . denote by @xmath118 the unique vertex in @xmath17 such that @xmath119 , and by @xmath120 the neighbor of @xmath117 distinct from @xmath118 and @xmath65 . create a new network @xmath121 by deleting @xmath65 and the edge @xmath122 from @xmath17 , splitting @xmath117 into one new source @xmath123 and one new sink @xmath124 ( so that either @xmath125 or @xmath126 in the counterclockwise order ) and replacing the edges @xmath127 and @xmath128 by new edges @xmath129 and @xmath130 in the same way as in the previous case , see figure [ fig : blackfork ] . let us classify the paths from @xmath65 to @xmath66 according to the number of times they traverse @xmath131 . similarly to the previous case , the total weight of the paths traversing @xmath131 only once is given by @xmath132 , where @xmath133 means that the measurement is taken in @xmath121 . any path @xmath41 traversing @xmath131 exactly twice can be represented as @xmath134 for some path @xmath135 from @xmath136 to @xmath137 in @xmath121 and a path @xmath138 from @xmath136 to @xmath66 in @xmath121 . clearly , @xmath139 is a cycle in @xmath42 , and @xmath140 . besides , @xmath141 for @xmath142 . therefore , by ( [ offcycle ] ) , @xmath143 . taking into account that @xmath144 and @xmath145 , we see that the total contribution of all paths traversing @xmath131 exactly twice to @xmath87 equals @xmath146 in general , the total contribution of all paths traversing @xmath131 exactly @xmath147 times to @xmath87 equals @xmath148 therefore , we find @xmath149 and the claim follows by induction , since the number of internal vertices in @xmath121 is @xmath116 . ( upper part ) and @xmath150 ( lower part),height=302 ] boundary measurements can be organized into a @xmath151 _ boundary measurement matrix _ @xmath152 in the following way . let @xmath153 and @xmath154 . we define @xmath155 , @xmath156 $ ] , @xmath157 $ ] , where @xmath158 . each network @xmath42 defines a rational map @xmath159 given by @xmath152 and called the _ boundary measurement map _ corresponding to @xmath42 . here and below @xmath160 denotes the space of @xmath151 matrices . [ ex : bmmex]consider the network shown in figure [ fig : pergraph ] . the corresponding boundary measurement matrix is a @xmath161 matrix given by @xmath162 a natural operation on networks is their _ concatenation _ , which consists , roughly speaking , in gluing some sinks / sources of one network to some of the sources / sinks of the other . we expect any poisson structure associated with networks to behave naturally under concatenation . to obtain from two planar networks a new one by concatenation , one needs to select a segment from the boundary of each disk and identify these segments via a homeomorphism in such a way that every sink ( resp . source ) contained in the selected segment of the first network is glued to a source ( resp . sink ) of the second network . we can then erase the common piece of the boundary along which the gluing was performed and identify every pair of glued edges in the resulting network with a single edge of the same orientation and with the weight equal to the product of two weights assigned to the two edges that were glued . as an illustration , let us review a particular but important case , in which sources and sinks of the network do not interlace . in this case , it is more convenient to view the network as located in a square rather than in a disk , with all sources located on the left side and sinks on the right side of the square . it will be also handy to label sources ( resp . sinks ) @xmath164 to @xmath1 ( resp . @xmath164 to @xmath165 ) going from the bottom to the top . this results in a different way of recording boundary measurements into a matrix . namely , if @xmath166 is the boundary measurements matrix we defined earlier , then now we associate with the network the matrix @xmath167 , where @xmath168 is the matrix of the longest permutation @xmath169 . we can now concatenate two networks of this kind , one with @xmath1 sources and @xmath165 sinks and another with @xmath165 sources and @xmath55 sinks , by gluing the sinks of the former to the sources of the latter . if @xmath170 , @xmath171 are @xmath151 and @xmath172 matrices associated with the two networks , then the matrix associated with their concatenation is @xmath173 . note that this `` visualization '' of the matrix multiplication is particularly relevant when one deals with factorization of matrices into products of elementary bidiagonal matrices . indeed , a @xmath174 diagonal matrix @xmath175 and elementary bidiagonal matrices @xmath176 and @xmath177 correspond to planar networks shown in figure [ fig : factor ] a , b and c , respectively ; all weights not shown explicitly are equal to 1 . the construction of poisson structures we are about to present is motivated by the way in which the network representation of the bidiagonal factorization reflects the _ standard poisson - lie structure _ on @xmath178 . we need to recall some facts about poisson - lie groups ( see , e.g.@xcite ) . let @xmath179 be a lie group equipped with a poisson bracket @xmath180 . @xmath179 is called a _ poisson - lie group _ if the multiplication map @xmath181 is poisson . perhaps , the most important class of poisson - lie groups is the one associated with classical r - matrices . let @xmath182 be a lie algebra of @xmath179 . assume that @xmath182 is equipped with a nondegenerate invariant bilinear form @xmath183 . an element @xmath184 is a _ classical r - matrix _ if it is a skew - symmetric operator that satisfies the _ modified classical yang - baxter equation _ ( mcybe ) @xmath185 - r \left ( [ r(\xi ) , \eta]\ + \ [ \xi , r(\eta ) ] \right ) = - [ \xi,\eta]\ . \label{mcybe}\ ] ] given a classical r - matrix @xmath186 , @xmath179 can be endowed with a poisson - lie structure as follows . let @xmath187 be the right and the left gradients for a function @xmath188 : @xmath189 then the bracket given by @xmath190 is a poisson - lie bracket on @xmath179 called the _ sklyanin bracket_. we are interested in the case @xmath191 and @xmath192 equipped with the trace - form @xmath193 then the right and left gradients ( [ rightleftgrad ] ) are @xmath194 where @xmath195 and the sklyanin bracket becomes @xmath196 every @xmath197 can be uniquely decomposed as @xmath198 where @xmath199 and @xmath200 are strictly upper and lower triangular and @xmath201 is diagonal . the simplest classical r - matrix on @xmath202 is given by @xmath203 substituting into ( [ sklyasln ] ) @xmath204 we obtain , for matrix entries @xmath205 , @xmath206 the bracket ( [ sklyasln ] ) extends naturally to a poisson bracket on the space @xmath207 of @xmath174 matrices . for example , the standard poisson - lie structure on @xmath208 is described by the relations @xmath209 which , when restricted to upper and lower borel subgroups of @xmath210 @xmath211 have an especially simple form @xmath212 the latter poisson brackets can be used to give an alternative characterization of the standard poisson - lie structure on @xmath178 . namely , define the canonical embedding @xmath213 ( @xmath214 $ ] ) that maps @xmath215 into @xmath178 as a diagonal @xmath161 block occupying rows and columns @xmath216 and @xmath217 . then the standard poisson - lie structure on @xmath178 is defined uniquely ( up to a scalar multiple ) by the requirement that restrictions of @xmath218 to @xmath219 are poisson . note that the network that represents @xmath220 looks like the second network in the figure above with the weights @xmath221 and @xmath222 attached to edges @xmath223 , @xmath224 and @xmath225 resp . , while the network that represents @xmath226 looks like the third network in the figure above with the weights @xmath221 and @xmath222 attached to edges @xmath227 , @xmath228 and @xmath229 resp . concatenation of several networks @xmath230 , @xmath231 , of these two kinds , with appropriately chosen order and with each diagram having its own pair of nontrivial weights @xmath232 , describes a generic element of @xmath178 ( see , e.g. @xcite ) . an example of such a network is given by figure [ fig : genfactor ] . on the other hand , due to the poisson - lie property , the poisson structure on @xmath178 is inherited from simple poisson brackets for parameters @xmath233 , which can be described completely in terms of networks : ( i ) the bracket of any two parameters is equal to their product times a constant ; ( ii ) this constant is equal to zero unless the corresponding edges have a common source / sink ; ( iii ) the constant equals @xmath234 if the corresponding edges follow one another around the source / sink in the counterclockwise direction . the corresponding poisson - lie structure on @xmath0 is obtained by requiring the determinant to be a casimir function . this example motivates conditions we impose below on a natural poisson structure associated with a 3-valent planar directed network . let g be a directed planar graph in a disk as described in section [ pandw ] . a pair @xmath235 is called a _ flag _ if @xmath53 is an endpoint of @xmath236 . to each internal vertex @xmath53 of @xmath17 we assign a 3-dimensional space @xmath237 with coordinates @xmath238 . we equip each @xmath237 with a poisson bracket @xmath239 . it is convenient to assume that the flags involving @xmath53 are labelled by the coordinates , as shown in figure [ fig : trefoils ] . , height=56 ] besides , to each boundary vertex @xmath66 of @xmath17 we assign a 1-dimensional space @xmath240 with the coordinate @xmath241 ( in accordance with the above convention , this coordinate labels the unique flag involving @xmath66 ) . define @xmath242 to be the direct sum of all the above spaces ; thus , the dimension of @xmath242 equals twice the number of edges in @xmath17 . note that @xmath242 is equipped with a poisson bracket @xmath243 , which is defined as the direct sum of the brackets @xmath239 ; that is , @xmath244 whenever @xmath245 and @xmath246 are not defined on the same @xmath237 . we say that the bracket @xmath243 is _ universal _ if each of @xmath239 depends only on the color of the vertex @xmath53 . define the weights @xmath28 by @xmath247 provided the flag @xmath235 is labelled by @xmath248 and the flag @xmath249 is labelled by @xmath250 . in other words , the weight of an edge is defined as the product of the weights of the two flags involving this edge . therefore , in this case the space of edge weights @xmath251 coincides with the entire @xmath252 , and the weights define a _ weight map _ @xmath253 . we require the pushforward of @xmath243 to @xmath252 by the weight map to be a well defined poisson bracket ; this can be regarded as an analog of the poisson lie property for groups . [ 6param ] universal poisson brackets @xmath243 such that the weight map @xmath26 is poisson form a @xmath254-parametric family defined by relations @xmath255 , i\ne j,\ ] ] at each white vertex @xmath53 and @xmath256 , i\ne j,\ ] ] at each black vertex @xmath53 . indeed , let @xmath53 be a white vertex , and let @xmath257 and @xmath258 be the two outcoming edges . by definition , there exist @xmath259 $ ] , @xmath49 , such that @xmath260 , @xmath261 . therefore , @xmath262 where @xmath263 stands for the pushforward of @xmath243 . recall that the poisson bracket in @xmath237 depends only on @xmath264 , @xmath265 and @xmath266 . hence the only possibility for the right hand side of the above relation to be a function of @xmath28 and @xmath267 occurs when @xmath268 , as required . black vertices are treated in the same way . let @xmath53 be a white vertex . a _ local gauge transformation _ at @xmath53 is a transformation @xmath269 defined by @xmath270 , where @xmath271 is a laurent monomial in @xmath264 , @xmath265 , @xmath266 . a local gauge transformation at a black vertex is defined by the same formulas , with @xmath271 replaced by @xmath272 . a _ global gauge transformation _ @xmath273 is defined by applying a local gauge transformation @xmath271 at each vertex @xmath53 . the composition map @xmath274 defines a network @xmath275 ; the graph of @xmath275 coincides with the graph of @xmath42 , and the weight @xmath276 of an edge @xmath277 is given by @xmath278 . therefore , the weights of the same path in @xmath42 and @xmath275 coincide . it follows immediately that @xmath279 provided both sides of the equality are well defined . our next goal is to look at poisson properties of the boundary measurement map . fix an arbitrary partition @xmath280 $ ] , @xmath281 , and let @xmath282 , @xmath283 . let @xmath284 stand for the set of all perfect planar networks in a disk with the sources @xmath65 , @xmath85 , sinks @xmath66 , @xmath86 , and edge weights @xmath28 defined by ( [ connect ] ) . we assume that the space of edge weights @xmath285 is equipped with the poisson bracket @xmath263 obtained as the pushforward of the 6-parametric family @xmath243 described in proposition [ 6param ] . [ psm ] there exists a @xmath102-parametric family of poisson brackets on @xmath163 with the following property : for any choice of parameters @xmath286 , @xmath287 in , this family contains a unique poisson bracket on @xmath163 such that for any network @xmath288 the map @xmath289 is poisson . relation ( [ gaugcom ] ) suggests that one may use global gauge transformations in order to decrease the number of parameters in the universal 6-parametric family described in proposition [ 6param ] . indeed , for any white vertex @xmath53 we consider a local gauge transformation @xmath290 with @xmath291 . evidently , @xmath292 with @xmath293 similarly , for each black vertex @xmath53 we consider a local gauge transformation @xmath294 with @xmath295 . evidently , @xmath296 with @xmath297 from now on we consider the 2-parametric family ( [ 2parw ] ) , ( [ 2parb ] ) instead of the 6-parametric family ( [ 6parw ] ) , ( [ 6parb ] ) . to define a poisson bracket on @xmath163 , it suffices to calculate the bracket for any pair of matrix entries and to extend it further via bilinearity and the leibnitz identity . to do this we will need the following two auxiliary functions : for any @xmath298 , @xmath299 define @xmath300 and @xmath301 note that both @xmath302 and @xmath303 are skew - symmetric : @xmath304 for any @xmath298 , @xmath305 . quadruples @xmath306 such that at least one of the values @xmath307 and @xmath308 is distinct from zero are shown in figure [ fig : twofunctions ] . for a better visualisation , pairs @xmath309 and @xmath310 are joined by a directed edge ; these edges should not be mistaken for edges of @xmath42 . and @xmath303,height=207 ] theorem [ psm ] is proved by presenting an explicit formula for the bracket on @xmath163 . [ psme ] the @xmath102-parametric family of poisson brackets on @xmath163 satisfying the conditions of theorem [ psm ] is given by @xmath311 where @xmath312 and @xmath313 satisfy , , @xmath314 $ ] , @xmath315 $ ] . first of all , let us check that that relations ( [ psme ] ) indeed define a poisson bracket on @xmath163 . since bilinearity and the leibnitz identity are built - in in the definition , and skew symmetry follows immediately from ( [ spar ] ) and ( [ scross ] ) , it remains to check the jacobi identity . [ jacobi ] the bracket @xmath316 satisfies the jacobi identity . the claim can be verified easily when at least one of the following five conditions holds true : @xmath317 ; @xmath318 ; @xmath319 and @xmath320 ; @xmath321 and @xmath322 ; @xmath323 and @xmath324 . in what follows we assume that none of these conditions holds . a simple computation shows that under this assumption , the jacobi identity for @xmath316 is implied by the following three identities for the functions @xmath302 and @xmath303 : for any @xmath325 , @xmath326 , @xmath327 the first identity in ( [ threeid ] ) is evident , since by ( [ scross ] ) , the first term is canceled by the fifth one , the second term is canceled by the sixth one , and the third term is canceled by the fourth one . to prove the second identity , assume to the contrary that there exist @xmath325 , @xmath326 such that the left hand side does not vanish . consequently , at least one of the three terms in the left hand side does not vanish ; without loss of generality we may assume that it is the first term . since @xmath328 , we get either @xmath329 or @xmath330 assume that ( [ order1 ] ) holds , than we have five possibilities for @xmath331 : \(i ) @xmath332 ; \(ii ) @xmath322 ; \(iii ) @xmath333 ; \(iv ) @xmath334 ; \(v ) @xmath320 . in cases ( i)-(iii ) condition @xmath335 implies @xmath336 . therefore , in case ( i ) we get @xmath337 provided @xmath338 , and @xmath339 provided @xmath340 . in both situations the second identity in ( [ threeid ] ) follows immediately . in case ( ii ) we get @xmath341 and @xmath342 in both situations the second identity in ( [ threeid ] ) follows immediately . in case ( iii ) we get @xmath343 provided @xmath344 , and @xmath345 provided @xmath346 . in both situations the second identity in ( [ threeid ] ) follows immediately . case ( iv ) , in its turn , falls into three cases depending on the location of @xmath216 ; these three cases are parallel to the cases ( i)-(iii ) above and are treated in the same way . finally , in case ( v ) we have to distinguish two subcases : @xmath347 and @xmath348 . in the first subcase we have @xmath349 while in the second subcase , @xmath350 and @xmath351 in both situations the second identity in ( [ threeid ] ) follows immediately . if the points are ordered counterclockwise as prescribed by ( [ order2 ] ) , the proof is very similar , with @xmath352 and @xmath353 replaced by @xmath354 and @xmath355 , correspondingly . to prove the third identity in ( [ threeid ] ) , assume to the contrary that there exist @xmath325 , @xmath326 such that the left hand side does not vanish consequently , @xmath356 , and hence once again one of ( [ order1 ] ) and ( [ order2 ] ) holds . , height=94 ] denote by @xmath357 the sum in the left hand side of the identity . if @xmath323 or @xmath324 then @xmath357 vanishes due to the skew - symmetry of @xmath303 . the remaining case amounts to checking all possible ways to insert a chord @xmath358 in the configurations presented in figure [ fig : btfly ] . the check itself in each case is trivial . for example , if the inserted chord is as shown by the dotted line in the left part of figure [ fig : btfly ] , then the last two terms of @xmath357 vanish , while the first two terms have opposite signs and absolute value 1 . if the inserted chord is as shown by a dotted line in the right part of figure [ fig : btfly ] , then one of the last two terms in @xmath357 vanishes and one of the first two terms has absolute value 1 . the two remaining terms have absolute value @xmath359 ; besides , they have the same sign , which is opposite to the sign of the term with absolute value 1 . other cases are similar and left to the reader . to complete the proof of theorem [ psme ] , it remains to check that @xmath360 for any pair of matrix entries @xmath361 and @xmath362 . the proof is similar to the proof of proposition [ sfree ] and relies on the induction on the number of inner vertices in @xmath363 . assume first that @xmath42 does not have inner vertices , and hence each edge of @xmath42 connects two boundary vertices . it is easy to see that in this case the poisson bracket computed in @xmath252 vanishes identically . let us prove that the bracket given by ( [ psme ] ) vanishes as well . we start with the case when both @xmath364 and @xmath365 are edges in @xmath42 . then @xmath366 and @xmath367 , since there is only one edge incident to each boundary vertex . therefore @xmath368 , and the first term in the right hand side of ( [ psme ] ) vanishes . besides , since @xmath42 is planar , @xmath369 , and the second term vanishes as well . next , let @xmath364 be an edge of @xmath42 , and @xmath365 be a non - edge . then @xmath370 , and hence both terms in the right hand side of ( [ psme ] ) vanish . finally , let both @xmath364 and @xmath365 be non - edges . then @xmath371 , and the second term in the right hand side of ( [ psme ] ) vanishes . the first term can be distinct from zero only if both @xmath372 and @xmath373 are edges in @xmath42 . once again we use planarity of @xmath42 to see that in this case @xmath374 , and hence the right hand side of ( [ psme ] ) vanishes . now we may assume that @xmath42 has @xmath98 inner vertices , and that ( [ twops ] ) is true for all networks with at most @xmath116 inner vertices and any number of boundary vertices . consider the unique neighbor of @xmath375 in @xmath42 . if this neighbor is another boundary vertex then the same reasoning as above applies to show that @xmath376 vanishes identically for any choice of @xmath377 , @xmath378 , which agrees with the behavior of @xmath263 . assume that the only neighbor of @xmath375 is a white inner vertex @xmath53 . define @xmath379 which corresponds to a network @xmath100 obtained from @xmath42 by deleting vertices @xmath375 , @xmath53 and the edge @xmath380 from @xmath17 and adding two new sources @xmath381 and @xmath382 so that @xmath383 , see figure [ fig : whitefork ] . let @xmath384 be the coordinates in @xmath237 after the local gauge transformation at @xmath53 , so that @xmath385 . then @xmath386 for any @xmath86 and any @xmath387 . since @xmath100 has @xmath116 inner vertices , @xmath388 satisfy relations similar to ( [ twops ] ) with @xmath42 replaced by @xmath100 and @xmath389 replaced by @xmath390 . relations ( [ twops ] ) follow immediately from this fact and ( [ simplerecalc ] ) , provided @xmath387 . in the latter case we have @xmath391 the first term in the right hand side of the expression above equals @xmath392 since @xmath393 ( the sign is negative if @xmath394 and positive if @xmath395 ) , the first term equals @xmath396 . similarly , the second term equals @xmath397 the fourth term equals @xmath398 and the sixths term equals @xmath399 with the same sign rule as for the first term . we thus see that @xmath400 since @xmath401 . assume now that the only neighbor of @xmath375 is a black inner vertex @xmath117 . define @xmath402 which corresponds to a network @xmath121 obtained from @xmath42 by deleting vertices @xmath375 , @xmath117 and the edge @xmath403 from @xmath17 and adding a new source @xmath136 and a new sink @xmath137 so that either @xmath404 , or @xmath404 , see figure [ fig : blackfork ] . let @xmath384 be the coordinates in @xmath405 after the local gauge transformation at @xmath117 , so that @xmath406 . [ recalc ] boundary measurements in the networks @xmath42 and @xmath121 are related by @xmath407 in the second formula above , sign @xmath408 corresponds to the cases @xmath409 or @xmath410 and sign @xmath411 corresponds to the cases @xmath412 or @xmath413 the first formula above was , in fact , already obtained in the proof of proposition [ sfree ] ; one has to take into account that after the local gauge transformation at @xmath117 we get @xmath414 , @xmath415 and @xmath416 . to get the second formula , we apply the same reasoning as in the proof of proposition [ sfree ] . the paths from @xmath417 to @xmath66 in @xmath42 are classified according to the number of times they traverse the edge @xmath131 . the total contribution of the paths not traversing @xmath131 at all to @xmath418 equals @xmath419 . each path @xmath41 that traverses @xmath131 exactly once can be decomposed as @xmath420 so that @xmath421 is a path from @xmath417 to @xmath124 in @xmath121 and @xmath422 is a path from @xmath123 to @xmath66 in @xmath121 . define @xmath423 , where @xmath236 is the edge between @xmath137 and @xmath136 belonging to the convex hull of the boundary vertices of @xmath121 ( see figure [ fig : recalc ] ) . clearly , @xmath424 . , height=453 ] assume first that @xmath425 which corresponds to the upper part of figure [ fig : recalc ] . then @xmath426 where @xmath427 is the contribution of all vertices of @xmath428 , including @xmath417 and @xmath124 , @xmath429 is the contribution of two consecutive edges of the convex hull of boundary vertices of @xmath121 calculated at @xmath123 , and @xmath430 is the total contribution calculated at the vertices of the convex hull lying between @xmath123 and @xmath417 . similarly , @xmath431 where @xmath432 is the contribution of all vertices of @xmath433 , including @xmath123 and @xmath434 , @xmath435 is the contribution of two consecutive edges of the convex hull of boundary vertices of @xmath121 calculated at @xmath124 , and @xmath436 is the total contribution calculated at the vertices of the convex hull lying between @xmath434 and @xmath437 . finally , @xmath438 and so @xmath439 . therefore , in this case @xmath440 , and the contribution of all such paths to @xmath418 equals @xmath441 . each additional traversing of the edge @xmath131 results in multiplying this expression by @xmath442 ; the proof of this fact is similar to the proof of proposition [ sfree ] . summing up we get the second formula with the @xmath411 sign , as desired . assume now that @xmath443 which corresponds to the lower part of figure [ fig : recalc ] . then @xmath444 where @xmath427 and @xmath429 are as in the previous case , @xmath445 is the contribution of two consecutive edges of the convex hull of boundary vertices of @xmath121 calculated at @xmath66 , @xmath430 is the total contribution calculated at the vertices of the convex hull lying between @xmath123 and @xmath434 , and @xmath446 is the total contribution calculated at the vertices of the convex hull lying between @xmath66 and @xmath417 . similarly , @xmath447 where @xmath432 and @xmath435 are as in the previous case , @xmath448 is the contribution of two consecutive edges of the convex hull of boundary vertices of @xmath121 calculated at @xmath417 , and @xmath436 is the total contribution calculated at the vertices of the convex hull lying between @xmath417 and @xmath124 . finally , @xmath449 , and so @xmath450 , since @xmath451 therefore , in this case @xmath452 , and the contribution of all such paths to @xmath418 equals @xmath453 . each additional traversing of @xmath131 once again results in multiplying this expression by @xmath442 . summing up we get the second formula with the @xmath408 sign , as desired . to treat the remaining two cases one makes use of remark [ dualdir ] and applies the same reasoning . since @xmath121 has @xmath116 inner vertices , @xmath454 satisfy relations similar to ( [ twops ] ) with @xmath42 replaced by @xmath121 , @xmath389 replaced by @xmath455 and @xmath16 replaced by @xmath456 . relations ( [ twops ] ) follow from this fact and lemma [ recalc ] via simple though tedious computations . for example , let @xmath457 . then the left hand side of ( [ twops ] ) is given by @xmath458 and the right hand side is given by @xmath459 treating @xmath460 , @xmath461 and @xmath462 as independent variables and equating coefficients of the same monomials in the above two expressions we arrive to the following identities : @xmath463 the latter can be checked easily by considering separately the following cases : @xmath464 ; @xmath465 ; @xmath466 ; @xmath467 . in each one of these cases all the functions involved in the above identities take constant values . let us recall the construction of poisson brackets on the grassmannian from @xcite . let @xmath3 be the grassmannian of @xmath1-dimensional subspaces in @xmath468 . given a @xmath1-element subset @xmath389 of @xmath469=\{1,\dots , n\}$ ] , the plcker coordinate @xmath470 is a function on the set of @xmath471 matrices which is equal to the value of the minor formed by the columns of the matrix indexed by the elements of @xmath389 . in what follows , we use notation @xmath472 for @xmath473 $ ] and @xmath474 $ ] . let @xmath475 be the open cell in @xmath3 characterized by non - vanishing of the plcker coordinate @xmath476}$ ] . elements of @xmath475 are parametrized by @xmath477 matrices in the following way : if @xmath478 admits a factorization into block - triangular matrices @xmath479 then @xmath480 represents an element of the cell @xmath475 . it is easy to check that plcker coordinates @xmath470 , @xmath481 of an element of @xmath475 represented by @xmath482 $ ] and minors @xmath483 of @xmath484 are related via @xmath485\setminus\{\alpha_1 , \ldots , \alpha_l\ } ) \cup \{\beta_1+k , \ldots , \beta_l+k\}}}{x_{[1,k]}}.\ ] ] note that , if the row index set @xmath486 in the above formula is contiguous then the sign in the right hand side can be expressed as @xmath487 . a poisson bracket @xmath180 on @xmath475 is defined via the relation @xmath488 in terms of matrix elements @xmath489 of @xmath484 , this bracket looks as follows : @xmath490 let us take a closer look at the 2-parameter family of poisson brackets obtained in theorem [ psme ] in the case when vertices @xmath492 on the boundary of the disk are sources and vertices @xmath493 are sinks , that is , when @xmath494 $ ] and @xmath495 $ ] . to simplify notation , in this situation we will write @xmath496 and @xmath497 instead of @xmath498,[k+1,n]}$ ] and @xmath499,[k+1,n]}$ ] . therefore , formula ( [ psme ] ) can be re - written as @xmath500 where @xmath501 corresponds to the boundary measurement between @xmath65 and @xmath502 . the first term in the equation above coincides ( up to a multiple ) with ( [ bracky ] ) . this suggests that it makes sense to investigate poisson properties of the boundary measurement map viewed as a map into @xmath3 , which will be the goal of this section . first , however , we will go back to the example , considered in sect . [ concat ] , where we associated with a network @xmath503 a matrix @xmath504 . written in terms of matrix entries @xmath505 of @xmath506 , bracket ( [ mbrack12 ] ) becomes @xmath507 if @xmath508 are matrices that correspond to networks @xmath509 then their product @xmath510 corresponds to the concatenation of @xmath511 and @xmath512 . this fact combined with theorem [ psm ] implies that the bracket ( [ abrack12 ] ) possesses the poisson - lie property . in fact , ( [ abrack12 ] ) is the sklyanin bracket ( [ sklyasln ] ) on @xmath491 associated with a deformation of the standard r - matrix ( [ standardr ] ) . indeed , it is known ( see @xcite ) that if @xmath513 is the standard r - matrix , @xmath357 is any linear operator on the set of diagonal matrices that is skew - symmetric w.r.t . the trace - form , and @xmath514 is the natural projection onto a subspace of diagonal matrices , then for any scalar @xmath515 , @xmath516 the linear combination @xmath517 satisfies mcybe ( [ mcybe ] ) and thus gives rise to a sklyanin poisson - lie bracket . define @xmath357 by @xmath518 and put @xmath519 substituting coordinate functions @xmath505 , @xmath520 into expression ( [ sklyasln ] ) for the sklyanin poisson - lie bracket associated with the r - matrix @xmath521 , we recover equation ( [ abrack12 ] ) . to summarize , we obtained [ psgl ] for any network @xmath522 and any choice of parameters @xmath286 , @xmath287 in , , the map @xmath523 is poisson with respect to the sklyanin bracket associated with the r - matrix @xmath521 , where @xmath312 and @xmath313 satisfy relations and . let @xmath42 be a network with the sources @xmath65 , @xmath85 and sinks @xmath66 , @xmath86 . following @xcite , we are going to interpret the boundary measurement map as a map into the grassmannian @xmath3 . to this end , we extend @xmath152 to a @xmath471 matrix @xmath524 as follows : \(i ) the @xmath525 submatrix of @xmath524 formed by @xmath1 columns indexed by @xmath389 is the identity matrix @xmath526 ; \(ii ) for @xmath527 $ ] and @xmath528 , the @xmath529-entry of @xmath524 is @xmath530 where @xmath531 is the number of elements in @xmath389 lying strictly between @xmath532 and @xmath533 in the linear ordering ; note that the sign is selected in such a way that the minor @xmath534}^{i(i_p\to j)}$ ] coincides with @xmath535 . we will view @xmath524 as a matrix representative of an element @xmath536 . the corresponding rational map @xmath537 is called the _ grassmannian boundary measurement map_. for example , the network presented in figure [ fig : grbmm ] defines a map of @xmath538 to @xmath539 given by the matrix @xmath540 clearly , @xmath541 belongs to the cell @xmath542 . therefore , we can regard matrix entries @xmath543 as coordinate functions on @xmath544 and rewrite ( [ psme ] ) as follows : @xmath545 if @xmath494 $ ] , then the 2-parametric family ( [ mbrack12 ] ) of poisson brackets on @xmath546 is defined by ( [ mbrack12 ] ) . a computation identical to the one presented in @xcite ( see sect . [ recallgsv1 ] ) shows that ( [ mbrack12 ] ) is the pushforward to @xmath547 of the sklyanin poisson - lie bracket ( [ sklyasln ] ) on @xmath207 associated with the r - matrix @xmath521 . the following result says that the families @xmath316 on different cells @xmath548 can be glued together to form the unique 2-parametric family of poisson brackets on @xmath3 that makes all maps @xmath541 poisson . [ psgr ] ( i ) for any choice of parameters @xmath312 and @xmath313 there exists a unique poisson bracket @xmath4 on @xmath3 such that for any network @xmath42 with @xmath1 sources , @xmath2 sinks and weights defined by , the map @xmath549 is poisson provided the parameters @xmath286 and @xmath287 defining the bracket @xmath263 on @xmath252 satisfy relations and . \(ii ) for any @xmath14 $ ] , @xmath550 , and @xmath551\setminus i$ ] , the restriction of @xmath4 to the cell @xmath548 coincides with the bracket @xmath316 given by in coordinates @xmath543 . both statements follow immediately from theorems [ psm ] and [ psme ] , the fact that @xmath547 is an open dense subset in @xmath3 , and the following proposition . [ coincbr ] for any @xmath552 and any @xmath553 , @xmath554 it is convenient to rewrite ( [ mbrack12 ] ) as @xmath555 and to treat the brackets @xmath556 and @xmath557 separately for a suitable choice of parameters @xmath312 and @xmath313 . let @xmath494 $ ] and denote @xmath558}$ ] simply by @xmath559 . then two independent brackets are given by @xmath560 ( for @xmath561 ) that coincides ( up to a multiple ) with the poisson bracket ( [ bracky ] ) on @xmath475 and @xmath562 ( for @xmath563 ) . let us start with the first of the two brackets above . for a generic element @xmath564 in @xmath3 represented by a matrix @xmath565 , consider any two minors , @xmath566 and @xmath567 , with row sets @xmath568 , @xmath569 , @xmath570 $ ] and column sets @xmath571 , @xmath572 , @xmath573 $ ] . using considerations similar to those in the proof of lemma 3.2 in @xcite , we obtain from ( [ firstbrack ] ) @xmath574 plcker coordinates @xmath470 ( for @xmath14 $ ] , @xmath550 ) and minors @xmath575 are related via @xmath576\setminus \mu ) \cup \{\nu\}}}{x_{[1,k]}}.\ ] ] denote @xmath577 } } = ( -1)^ { l(l-1)/2 } \wx_{[1,k]\setminus i}^{i\setminus[1,k]},\ ] ] where @xmath578\setminus i \right \|$ ] . then ( [ firstbrackminors ] ) gives rise to poisson relations @xmath579 } \sum_{i'\in i'\cap [ 1,k ] } \s ( i - i')a_{i(i\to i ' ) } a_{i'(i'\to i)}\\ & + \sum_{i\in i\setminus [ 1,k ] } \sum_{i'\in i'\setminus [ 1,k ] } \s ( i - i ' ) a_{i(i\to i ' ) } a_{i'(i'\to i)}\\ \label{firstbrackplu } & = \sum_{i\in i } \sum_{i'\in i ' } \varepsilon_{i i ' } a_{i(i\to i ' ) } a_{i'(i'\to i)},\end{aligned}\ ] ] where @xmath580 let us fix a @xmath1-element index set @xmath581 and use ( [ firstbrackplu ] ) to compute , for any @xmath582 $ ] and @xmath583 , the bracket @xmath584 . taking into account that @xmath585 $ , $ j \in j$},\ ] ] we get @xmath586 where in the third step we have used the short plcker relation . relation ( [ coincbr ] ) for the first bracket follows from for any @xmath552 and any @xmath587 , @xmath588 denote the left hand side of ( [ newspar ] ) by @xmath589 . let us prove that this expression depends only on the counterclockwise order of the numbers @xmath590 , @xmath331 , @xmath591 , @xmath592 and does not depend on the numbers themselves . assume first that all four numbers are distinct . in this case @xmath589 is invariant with respect to the cyclic shift of the variables , hence it suffices to verify the identity @xmath593 provided the counterclockwise orders of @xmath590 , @xmath331 , @xmath591 , @xmath592 and @xmath590 , @xmath594 , @xmath591 , @xmath592 coincide . this is trivial unless @xmath595 or @xmath596 , since all four summands retain their values . in the remaining cases the second and the fourth summands retain their values , and we have to check the identities @xmath597 and @xmath598 . the first of them follows from the identity @xmath599 for @xmath600 , and the second one , from the identity @xmath601 for @xmath602 . if @xmath603 or @xmath604 ( other coincidences are impossible since @xmath552 , @xmath587 ) , @xmath605 degenerates to a function of three variables , which is again invariant with respect to cyclic shifts , therefore all the above argument applies as well . to obtain ( [ newspar ] ) it remains to check that @xmath606 , which can be done separately in all the cases mentioned in ( [ spar ] ) . next , let us turn to the poisson bracket ( [ secondbrack ] ) . for any two monomials in matrix entries of @xmath524 , @xmath607 , and @xmath608 , we have @xmath609 observe that the double sum above is invariant under any permutation of indices within sets @xmath568 , @xmath569 , @xmath571 , @xmath572 . this means that for minors @xmath610 we have @xmath611 and the resulting poisson relations for functions @xmath612 are @xmath613 \setminus i } \sum_{i'\in [ 1,k ] \setminus i ' } \s ( i - i')\right.\\ & \qquad\qquad\qquad + \left . \sum_{i\in i\setminus [ 1,k ] } \sum_{i'\in i'\setminus [ 1,k ] } \s ( i - i ' ) \right)\\ & = a_i a_{i'}\sum_{i\in i } \sum_{i'\in i ' } \varepsilon_{i i'},\end{aligned}\ ] ] where @xmath614 has the same meaning as above . these relations imply @xmath615 relation ( [ coincbr ] ) for the second bracket follows now from for any @xmath552 and any @xmath587 , @xmath616 the proof of ( [ newscross ] ) is similar to the proof of ( [ newspar ] ) . this proves proposition [ coincbr ] , and hence theorem [ psgr ] . in this subsection we interpret the natural action of @xmath0 on @xmath3 in terms of planar networks . first , note that _ any _ element of @xmath0 can be represented by a planar network built by concatenation from building blocks ( elementary networks ) described in fig . [ fig : factor ] . to see this , one needs to observe that an elementary transposition matrix @xmath617 can be factored as @xmath618 which implies that any permutation matrix can be represented via concatenation of elementary networks . consequently , any elementary triangular matrices @xmath619 , @xmath620 , @xmath621 , can be factored as @xmath622 where @xmath623 is the permutation matrix that corresponds to the permutation @xmath624 the claim then follows from the bruhat decomposition and constructions presented in section [ concat ] . consider now a network @xmath363 and a network @xmath625 representing an element @xmath626 as explained above . we will concatenate @xmath42 and @xmath625 according to the following rule : in @xmath625 reverse directions of all horizontal paths @xmath627 for @xmath85 , changing every edge weight @xmath26 involved to @xmath628 , and then glue the left boundary of @xmath625 to the boundary of @xmath42 in such a way that boundary vertices with the same label are glued to each other . denote the resulting net by @xmath629 . let @xmath524 and @xmath630 be the signed boundary measurement matrices constructed according to the recipe outlined at the beginning of section [ indpoisson ] . matrices @xmath630 and @xmath631 are representatives of the same element in @xmath3 . to check that @xmath630 coincides with the result of the natural action of @xmath0 on @xmath3 induced by the right multiplication , it suffices to consider the case when @xmath506 is a diagonal or an elementary bidiagonal matrix , that is when @xmath625 is one of the elementary networks in fig . [ fig : factor ] . if @xmath625 is the first diagram in fig . [ fig : factor ] , then the boundary measurements in @xmath632 are given by @xmath633 , @xmath634 , where @xmath87 are the boundary measurements in @xmath42 . this is clearly consistent with the natural @xmath0 action . now let @xmath635 ( the case @xmath636 can be treated similarly ) . then @xmath637 ( in the last line above we used lemma [ recalc ] ) . recall that for @xmath86 , an entry @xmath638 of @xmath524 coincides with @xmath639 up to the sign @xmath640 . by the construction in section [ indpoisson ] , if @xmath641 , then @xmath642 for @xmath643 such that @xmath644 and for all @xmath86 . then the first line of ( [ m_a ] ) shows that @xmath645 , where the first elementary matrix is @xmath525 and the second one is @xmath174 . if @xmath646 , then @xmath647 for all @xmath156 $ ] , and the second line of ( [ m_a ] ) results in @xmath648 . the latter equality is also clearly valid for the case @xmath649 , @xmath650 . finally , if @xmath651 , @xmath652 , let @xmath653 $ ] be the index such that @xmath654 , and let @xmath655 denote the @xmath216th column of @xmath524 . then the @xmath525 submatrix of @xmath656 formed by the columns indexed by @xmath389 is @xmath657 , where @xmath658 with the only nonzero element in the @xmath98th position . @xmath659 is the representative of @xmath660\in g_k(n)$ ] that has an identity matrix as its @xmath525 submatrix formed by columns indexed by @xmath389 . we need to show that @xmath661 . first note that @xmath662 and so , for all @xmath86 , @xmath663 if @xmath664 , then @xmath665 thus , to see that @xmath666 , it is enough to check that the sign assignment that was used in lemma [ recalc ] is consistent with the formula @xmath667 this can be done by direct inspection and is left to the reader as an exercise . recall that if @xmath668 is a lie subgroup of a poisson - lie group @xmath179 , then a poisson structure on the homogeneous space @xmath669 is called _ poisson homogeneous _ if the action map @xmath670 is poisson . now that we have established that the natural right action of @xmath0 on @xmath3 can be realized via the operation on networks described above , theorems [ psme ] , [ psgl ] and [ psgr ] immediately imply the following [ poihomo ] for any choice of parameters @xmath671 , the grassmannian @xmath3 equipped with the bracket @xmath4 is a poisson homogeneous space for @xmath0 equipped with the `` matching '' poisson - lie bracket . it is not difficult to deduce theorem [ poihomo ] from the general theory of poisson homogeneous spaces ( see , e.g. @xcite ) . the proof above illustrates capabilities of the network approach . first , we recall the basics of cluster algebras of geometric type . the definition that we present below is not the most general one , see , e.g. , @xcite for a detailed exposition . the _ coefficient group _ @xmath672 is a free multiplicative abelian group of a finite rank @xmath165 with generators @xmath673 . an _ ambient field _ is the field @xmath674 of rational functions in @xmath8 independent variables with coefficients in the field of fractions of the integer group ring @xmath675 $ ] ( here we write @xmath676 instead of @xmath677 ) . a _ seed _ ( of _ geometric type _ ) in @xmath674 is a pair @xmath678 , where @xmath679 is a transcendence basis of @xmath674 over the field of fractions of @xmath680 and @xmath681 is an @xmath682 integer matrix whose principal part @xmath683 ( that is , the @xmath174 submatrix formed by the columns @xmath684 ) is skew - symmetric . the @xmath8-tuple @xmath685 is called a _ cluster _ , and its elements @xmath686 are called _ cluster variables_. denote @xmath687 for @xmath688 $ ] . we say that @xmath689 is an _ extended cluster _ , and @xmath690 are _ stable variables_. it is convenient to think of @xmath674 as of the field of rational functions in @xmath691 independent variables with rational coefficients . given a seed as above , the _ adjacent cluster _ in direction @xmath692 $ ] is defined by @xmath693 where the new cluster variable @xmath694 is given by the _ exchange relation _ @xmath695 here , as usual , the product over the empty set is assumed to be equal to @xmath164 . we say that @xmath696 is obtained from @xmath697 by a _ matrix mutation _ in direction @xmath1 and write @xmath698 if @xmath699 given a seed @xmath678 , we say that a seed @xmath700 is _ adjacent _ to @xmath701 ( in direction @xmath1 ) if @xmath702 is adjacent to @xmath685 in direction @xmath1 and @xmath703 . two seeds are _ mutation equivalent _ if they can be connected by a sequence of pairwise adjacent seeds . the _ cluster algebra _ ( of _ geometric type _ ) @xmath704 associated with @xmath701 is the @xmath680-subalgebra of @xmath674 generated by all cluster variables in all seeds mutation equivalent to @xmath701 . let @xmath180 be a poisson bracket on the ambient field @xmath674 . we say that it is _ compatible _ with the cluster algebra @xmath705 if , for any extended cluster @xmath689 , one has @xmath706 where @xmath707 are constants for all @xmath708 $ ] . the matrix @xmath709 is called the _ coefficient matrix _ of @xmath180 ( in the basis @xmath710 ) ; clearly , @xmath711 is skew - symmetric . consider , along with cluster and stable variables @xmath712 , another @xmath713-tuple of rational functions denoted @xmath714 and defined by @xmath715 where @xmath716 is an integer , @xmath717 for @xmath718 . we say that the entries @xmath719 $ ] form _ a @xmath720-cluster_. it is proved in @xcite , lemma 1.1 , that @xmath721 can be selected in such a way that the transformation @xmath722 is non - degenerate , provided @xmath723 . recall that a square matrix @xmath506 is _ reducible _ if there exists a permutation matrix @xmath41 such that @xmath724 is a block - diagonal matrix , and _ otherwise.the following result is a particular case of theorem 1.4 in @xcite . [ thm:1.4 ] assume that @xmath723 and the principal part of @xmath697 is irreducible . then a poisson bracket is compatible with @xmath725 if and only if its coefficient matrix in the basis @xmath720 has the following property : its @xmath726 submatrix formed by the first @xmath8 rows is proportional to @xmath697 . in @xcite we constructed the cluster algebra @xmath727 on the open cell @xmath475 in the grassmannian @xmath3 . this structure can be viewed as a restriction of the cluster algebra of the coordinate ring of @xmath3 described in @xcite using combinatorial properties of postnikov s construction . let us briefly review the construction of @xcite . for every @xmath728-entry of the matrix @xmath484 defined in ( [ defy ] ) , put @xmath729 and @xmath730}^{[j , j+l(i , j)]}\ .\ ] ] submatrices of @xmath484 whose determinants define functions @xmath731 are depicted in fig . [ fviay ] . , height=151 ] the initial extended cluster consists of functions @xmath732\setminus [ i - l(i , j),\ i ] ) \cup [ j+k , j+l(i , j)+k]}}{x_{[1,k]}},\\ i\in [ 1,k],\quad j\in [ 1 , m ] \ , \end{gathered}\ ] ] where @xmath470 denote plcker coordinates of the element of @xmath547 represented by the matrix @xmath733 $ ] . functions @xmath734 serve as stable coordinates . the entries of @xmath697 are all @xmath735 or @xmath736s . thus it is convenient to describe @xmath697 by a directed graph @xmath737 . the vertices of @xmath737 correspond to all columns of @xmath697 , and , since @xmath697 is rectangular , the corresponding edges are either between the cluster variables or between a cluster variable and a stable variable . in our case , @xmath737 is a directed graph with vertices forming a rectangular @xmath151 array and labeled by pairs of integers @xmath728 , and edges @xmath738 and @xmath739 ( cf . [ fig : graph ] ) . corresponding to @xmath475,height=132 ] the main goal of this section is to use networks in order to show that , in fact , every poisson structure in the 2-parameter family @xmath4 described in theorem [ psgr ] is compatible with @xmath727 . let @xmath20 be a perfect planar network in a disk . graph @xmath17 divides the disk into a finite number of connected components called _ faces_. the boundary of each face consists of edges of @xmath17 and , possibly , of several arcs bounding the disk . a face is called _ bounded _ if its boundary contains only edges of @xmath17 and _ unbounded _ otherwise . in this section we additionally require that each edge of @xmath17 belongs to a path from a source to a sink . this is a technical condition that ensures that the two faces separated by an edge are distinct . clearly , the edges that violate this condition do not influence the boundary measurement map and may be eliminated from the graph . given a face @xmath740 , we define its _ face weight _ @xmath741 as the laurent monomial in edge weights @xmath28 , @xmath22 , given by @xmath742 where @xmath743 if the direction of @xmath236 is compatible with the counterclockwise orientation of the boundary @xmath744 and @xmath745 otherwise . for example , the face weights for the network shown in figure [ fig : grbmm ] are @xmath746 , @xmath747 , @xmath748 , @xmath749 for four unbounded faces and @xmath750 for the only bounded face . similarly to the space of edge weights @xmath251 , we can define the _ space of face weights _ @xmath751 ; a point of this space is the graph @xmath17 as above with the faces weighted by real numbers obtained by specializing the variables @xmath19 in the expressions for @xmath28 to nonzero values and subsequent computation via ( [ faceweight ] ) ( recall that by definition , @xmath28 do not vanish on @xmath24 ) . by the euler formula , the number of faces of @xmath17 equals @xmath752 , and so @xmath751 is a semialgebraic subset in @xmath753 . in particular , the product of the face weights over all faces equals @xmath164 identically , since each edge enters the boundaries of exactly two faces in opposite directions . in general , the edge weights can not be restored from the face weights . however , the following proposition holds true . [ bmviay ] the weight @xmath754 of an arbitrary path @xmath41 between a source @xmath65 and a sink @xmath66 is a monomial in the face weights . assume first that all the edges in @xmath41 are distinct . extend @xmath41 to a cycle @xmath67 by adding the arc on the boundary of the disk between @xmath66 and @xmath65 in the counterclockwise direction . then the product of the face weights over all faces lying inside @xmath67 equals the product of the conductivities over the edges of @xmath41 , which is @xmath754 . similarly , for any simple cycle @xmath50 in @xmath17 , its weight @xmath755 equals the product of the face weights over all faces lying inside @xmath50 . it remains to use ( [ offcycle ] ) and to care that the cycles that are split off are simple . the latter can be guaranteed by the _ loop erasure _ procedure ( see @xcite ) that consists in traversing @xmath41 from @xmath65 to @xmath66 and splitting off the cycle that occurs when the first time the current edge of @xmath41 coincides with the edge traversed earlier . it follows from lemma [ bmviay ] that one can define boundary measurement maps @xmath756 and @xmath757 so that @xmath758 where @xmath759 is given by ( [ faceweight ] ) . our next goal is to write down the 2-parametric family of poisson structures induced on @xmath751 by the map @xmath246 . let @xmath20 be a perfect planar network . in this section it will be convenient to assume that boundary vertices are colored in _ gray_. define the _ directed dual network _ @xmath760 as follows . vertices of @xmath761 are the faces of @xmath42 . edges of @xmath761 correspond to the edges of @xmath42 with endpoints of different colors ; note that there might be several edges between the same pair of vertices in @xmath761 . an edge @xmath762 of @xmath761 corresponding to @xmath236 is directed in such a way that the white endpoint of @xmath236 ( if it exists ) lies to the left of @xmath762 and the black endpoint of @xmath236 ( if it exists ) lies to the right of @xmath236 . the weight @xmath763 equals @xmath461 if the endpoints of @xmath236 are white and black , @xmath312 if the endpoints of @xmath236 are white and gray and @xmath764 if the endpoints of @xmath236 are black and gray . an example of a directed planar network and its directed dual network is given in fig . [ fig : dualnet ] . [ psviay ] the @xmath102-parametric family @xmath765 is given by @xmath766 let @xmath277 be a directed edge . we say that the flag @xmath249 is _ positive _ , and the flag @xmath235 is _ negative_. the color of a flag is defined as the color of the vertex participating in the flag . let @xmath740 and @xmath767 be two faces of @xmath42 . we say that a flag @xmath235 is _ common _ to @xmath740 and @xmath767 if both @xmath53 and @xmath236 belong to @xmath768 . clearly , the bracket @xmath769 can be calculated as the sum of the contributions of all flags common to @xmath740 and @xmath767 . assume that @xmath235 is a positive white flag common to @xmath740 and @xmath767 , see fig . [ fig : comflag ] . then @xmath770 and @xmath771 , where @xmath772 are the weights of flags involving @xmath53 and @xmath773 , see section [ pssc ] . therefore , by ( [ 6parw ] ) , the contribution of @xmath235 equals @xmath774 , which by ( [ condw ] ) equals @xmath775 . assume now that @xmath235 is a negative white flag common to @xmath740 and @xmath767 , see fig . [ fig : comflag ] . in this case @xmath776 and @xmath771 , so the contribution of @xmath235 equals @xmath777 . in a similar way one proves that the contribution of a positive black flag common to @xmath740 and @xmath767 equals @xmath778 , and the contribution of a negative black flag common to @xmath740 and @xmath767 equals @xmath779 . finally , the contributions of positive and negative gray flags are clearly equal to zero . the statement of the lemma now follows from the definition of the directed dual network . now we can formulate the main theorem of this section . [ thm : poisscomp ] any poisson structure in the two - parameter family @xmath4 is compatible with the cluster algebra @xmath727 . by theorem [ thm:1.4 ] , it is enough to choose an initial extended cluster and to compare the coefficient matrix of @xmath4 in the basis @xmath720 with the exchange matrix for this cluster . to find the coefficient matrix of @xmath4 we define a special network @xmath780 with @xmath1 sources and @xmath165 sinks . the graph of @xmath780 has @xmath781 faces . each bounded face is a hexagon ; all bounded faces together form a @xmath782 parallelogram on the hexagonal lattice . edges of the hexagons are directed north , south - east and south - west . each vertex of degree 2 on the left boundary of this parallelogram is connected to a source , and each vertex of degree 2 on the upper boundary of this parallelogram is connected to a sink . the remaining vertices of degree 2 on the lower and the right boundaries of the parallelogram are eliminated : two edges @xmath783 are replaced by one edge @xmath784 and the intermediate vertex @xmath53 is deleted . the sources are labelled counterclockwise from 1 to @xmath1 , sinks from @xmath785 to @xmath8 . the faces are labelled by pairs @xmath786 such that @xmath787 $ ] , @xmath788 $ ] . the unbounded faces are labelled @xmath789 and @xmath790 except for one face , which is not labelled at all . the network @xmath780 is defined by assigning a face weight @xmath489 to each labelled face @xmath786 . consequently , the directed dual network @xmath791 forms the dual triangular lattice . all edges of @xmath791 incident to the vertices @xmath786 , @xmath792 $ ] , @xmath793 $ ] are of weight @xmath461 . for an example of the construction for @xmath794 , @xmath795 see fig . [ fig : postnet ] . the edge weights of the dual network that are not shown explicitly are equal to @xmath461 . and its directed dual @xmath796,height=188 ] as the first step of the proof , we express the cluster variables @xmath797 via the face weights of @xmath780 . [ pmfviay ] for any @xmath787 $ ] , @xmath798 $ ] , one has @xmath799 since the graph of @xmath780 is acyclic , one can use the lindstrm lemma @xcite to calculate the minors of the boundary measurement matrix . assume first that @xmath800 . by ( [ eq:3.16 ] ) and the lindstrm lemma , @xmath797 equals to the sum of the products of the path weights for all @xmath216-tuples of nonintersecting paths between the sources @xmath801 and the sinks @xmath802 , each product being taken with a certain sign . note that there exists a unique path from @xmath13 to @xmath803 in @xmath780 . after this path is chosen , there remains a unique path from @xmath29 to @xmath804 , and so on . moreover , a path from @xmath13 to any sink @xmath805 with @xmath806 cuts off the sink @xmath803 . therefore , there exists exactly one @xmath216-tuple of paths as required . relation ( [ eq : pmfviay ] ) for @xmath800 now follows from the proof of lemma [ bmviay ] . the case @xmath807 is similar to the above one and relays on the uniqueness of an @xmath808-tuple of nonintersecting paths between the sources @xmath809 and the sinks @xmath810 . here we start with the unique path from @xmath65 to @xmath11 , then choose the unique remaining path between @xmath811 and @xmath812 , and so on . the next step is the calculation of @xmath720-coordinates for the initial extended cluster ( [ eq:3.25 ] ) via face weights . [ tauviay1 ] the @xmath720-coordinates corresponding to the cluster variables of the initial extended cluster are given by @xmath813,\ ; j\in [ 2,m].\ ] ] combining the definition of @xmath720-coordinates via ( [ eq:1.3 ] ) and the description of the exchange matrix in the basis @xmath814 provided by fig . [ fig : graph ] , we conclude that @xmath815,\ ; j\in [ 2,m];\ ] ] here we assume that @xmath816 if @xmath817 or @xmath818 . next , by ( [ eq : pmfviay ] ) , @xmath819 for @xmath820 $ ] , @xmath821 $ ] , and the result follows . the operation that expresses @xmath720-coordinates in terms of face weights is an analog of the twist studied in @xcite . the expressions for the @xmath720-coordinates that correspond to the stable variables of the initial extended cluster ( [ eq:3.25 ] ) are somewhat cumbersome . recall that by ( [ eq:1.3 ] ) , each stable variable enters the expression for the corresponding @xmath720-coordinate with some integer exponent @xmath822 . let us denote @xmath823 for @xmath824 , @xmath798 $ ] and @xmath787 $ ] , @xmath595 . [ tauviay2 ] the @xmath720-coordinates corresponding to the stable variables of the initial extended cluster are given by @xmath825$},\\ \pm\displaystyle\prod_{q = k+2}^n\;\prod_{p = i}^{\min\{n , i - q+k+2\}}y_{pq } \qquad & \text{for $ i\in [ 2,k-1]$ , $ j=1$},\\ \pm\displaystyle\prod_{p=1}^{k-1}y_{pn } \qquad & \text{for $ i = k$ , $ j = m$},\\ \pm\displaystyle\prod_{q = k+2}^{n}y_{1q } \qquad & \text{for $ i=1 $ , $ j=1$},\\ \pm\displaystyle\prod_{p=1}^k\prod_{q = k+1}^n y_{pq}^{-\min\{k - p , q - k-1\ } } \qquad & \text{for $ i = k$ , $ j=1$}. \end{cases}\ ] ] it suffices to note that @xmath826$},\\ f_{i,2}/f_{i-1,2 } \qquad & \text{for $ i\in [ 2,k-1]$ , $ j=1$},\\ f_{k-1,m } \qquad & \text{for $ i = k$ , $ j = m$},\\ f_{12 } \qquad & \text{for $ i=1 $ , $ j=1$},\\ 1/f_{k-1,2 } \qquad & \text{for $ i = k$ , $ j = k+1 $ } , \end{cases}\ ] ] and to apply lemma [ pmfviay ] . to conclude the proof of the theorem we build the network @xmath827 representing the family of poisson brackets @xmath4 in the basis @xmath720 . the vertices of @xmath827 are @xmath720-coordinates @xmath828 , @xmath829 $ ] , @xmath798 $ ] . an edge from @xmath828 to @xmath830 with weight @xmath222 means that @xmath831 . it follows immediately from lemmas [ psviay ] and [ tauviay1 ] that the induced subnetwork @xmath832 of @xmath827 spanned by the vertices @xmath828 , @xmath820 $ ] , @xmath821 $ ] is isomorphic to the subnetwork @xmath833 of @xmath791 spanned by the vertices @xmath834 , @xmath835 $ ] , @xmath836 $ ] . we thus get an isomorphism between @xmath832 and the corresponding induced subgraph of @xmath737 , see fig . [ fig : iso ] . under this isomorphism each edge of weight @xmath461 is mapped to an edge of weight 1 . and @xmath737,height=151 ] for the remaining vertices of @xmath827 , that is , @xmath828 , @xmath787 $ ] , @xmath595 or @xmath824 , @xmath798 $ ] , we have to check the edges connecting them to the vertices of @xmath832 . consider for example the case @xmath824 , @xmath837 $ ] . then , by lemma [ tauviay2 ] , the bracket @xmath838 for @xmath839 is defined by the edges incident to the vertex subset of @xmath791 corresponding to the factors in the right hand side of the first formula in ( [ eq : tauviay2 ] ) , see fig . [ fig : contract ] . clearly , any vertex @xmath840 other than @xmath841 and @xmath842 is connected to this subset by an even number of edges ( more exactly , 0 , 2 , 4 or 6 ) , all of them of weight @xmath461 . since exactly half of the edges are directed to @xmath834 , the bracket between @xmath843 and @xmath844 vanishes by lemma [ psviay ] . the remaining two edges connecting the contracted set to the vertices @xmath841 and @xmath842 correspond to the two edges connecting the vertex @xmath845 to @xmath846 and @xmath847 in @xmath737 , see fig . [ fig : contract ] . , height=151 ] finally , factor @xmath848 commutes with all @xmath849 . indeed , given an edge between @xmath839 and @xmath850 , define its degree as the exponent of @xmath850 in expression ( [ eq : pmfviay ] ) for @xmath848 . we have to prove that the sum of the degrees of all edges entering @xmath830 equals the sum of the degrees of all edges leaving @xmath830 . this fact can be proved by analyzing all possible configurations of edges , see fig . [ fig : baldegree ] presenting these configurations up to reflection . and their degrees , height=113 ] other cases listed in lemma [ tauviay2 ] are treated in the same way . we wish to express gratitude to a. postnikov who explained to us the details of his construction and to s. fomin for stimulating discussions . m. g. was supported in part by nsf grant dms # 0400484 . m. s. was supported in part by nsf grants dms # 0401178 and phy#0555346 . authors were also supported by the bsf grant # 2002375 . a. v. is grateful to centre interfacultaire bernoulli at cole polytechnique fdrale de lausanne for hospitality during his spring 2008 visit . we are also grateful to the institute of advanced studies of the hebrew university of jerusalem for the invitation to the midrasha mathematicae in may 2008 , during which this paper was completed . j. moser , _ finitely many mass points on the line under the influence of the exponential potential - an integrable system_. dynamical systems , theory and applications , 467497 , lecture notes in physics * 38 * , springer , berlin , 1975 . a. reyman and m. semenov - tian - shansky _ group - theoretical methods in the theory of finite - dimensional integrable systems _ encyclopaedia of mathematical sciences , vol.16 , springer verlag , berlin , 1994 pp .	we investigate poisson properties of postnikov s map from the space of edge weights of a planar directed network into the grassmannian . we show that this map is poisson if the space of edge weights is equipped with a representative of a 6-parameter family of universal quadratic poisson brackets and the grasmannian is viewed as a poisson homogeneous space of the general linear group equipped with an appropriately chosen r - matrix poisson - lie structure . we also prove that poisson brackets on the grassmannian arising in this way are compatible with the natural cluster algebra structure .
the color of a galaxy is determined primarily by its star formation history ( sfh ) and the amount of dust attenuation present , with significant additional contributions from metallicity and dust geometry . an empirical relation between sfh , dust attenuation , and color is therefore a useful constraint on models of galaxy formation , and can aid in the interpretation of high redshift galaxy observations where measurements are difficult . here we investigate such an empirical relation , using a representative sample of galaxies observed from the ultraviolet to the infrared by _ galex _ , sdss , and _ spitzer_. the long wavelength coverage allows us to construct simple but robust measures of dust attenuation that are relatively free of a dependence on sfh . similarly , the sdss spectroscopy allows us to measure sfh diagnostics we use a 4000 break measure that are largely free of a dependence on dust attenuation . earlier studies of the effect of dust on galaxy colors have focused on the ultraviolet ( uv , @xmath6 ) colors of galaxies . a primary reason for this is that the intrinsic , underlying uv color ( before attenuation ) is relatively insensitive to the sfh of the galaxy , when compared to the effects of dust attenuation . also , studies of the highest redshift galaxies are often restricted to the restframe uv so that estimates of dust attenuation must be made using uv colors . @xcite derived the effective attenuation properties of dust in star - bursting galaxies by comparing the total dust absorption measured from the ratio of dust emission to uv emission to the change in uv color of a sample of starforming galaxies ( the so - called irx-@xmath5 relation ) . this locally derived relation has been used extensively to determine the dust attenuation in galaxies at higher redshift using their uv colors ( e.g. @xcite ) . however , more recent studies @xcite have shown that there is significant scatter in this relation , especially when less rapidly starforming galaxies are included . @xcite show , using stellar population synthesis models with significant recent bursts , that both sfh and dust are expected to affect the uv color of normal galaxies since at a constant value of dust attenuation galaxies with an older stellar population should appear redder in the uv because of their redder intrinsic uv spectra . they thus propose that a measure of sfh specifically the 4000 break can be used to explain the scatter in the irx-@xmath5 relation . we have constructed a sample of galaxies observed from the uv to the mid / far - infrared . our primary data set is the sample of local ( @xmath2 ) galaxies observed spectroscopically by sdss and analyzed by @xcite and @xcite ( hereafter sdss / mpa galaxies ) . the uv data is taken from pipeline processed _ galex _ observations of the lockman hole , with exposure times of @xmath7 ks . the pipeline - produced _ galex _ catalogs are searched for objects within 3 `` of the sdss / mpa galaxy locations , and the nearest object is taken as the match to the sdss galaxy . uv flux measurements are made in elliptical kron apertures . in the optical we use sdss petrosian magnitudes . the infrared data is provided by the swire _ spitzer _ observations of the lockman hole @xcite . we have performed aperture photometry in the swire team processed 3.6 through 7.8 irac images and 24 mips images at the location of each of the sdss / mpa galaxies , using a 7 '' radius aperture ( 12 " at 24 ) . the fluxes are then aperture corrected to total magnitudes . systematic errors in ir flux due to calibration uncertainty , aperture corrections , and the resolved nature of many of the sources amount to @xmath8 . the resulting uv through 3.6 magnitudes are @xmath9-corrected to @xmath10 ( e.g. @xmath11 , @xmath12 , etc . ) using the method of @xcite . at longer wavelengths dust emission becomes more important than stellar emission , and we use a different method to ` @xmath9-correct ' the data : we choose the best fitting redshifted @xcite model ir sed , on the basis of the observed 8 to 24 flux ratio . this sed is then normalized using the measured 24 flux , and the integrated far - infrared ( 8 - 1000 ) dust luminosities ( @xmath13 ) are derived . note that the different @xcite seds have @xmath14 ratios that are different by a factor of up to five . we have checked that our results would not change significantly if we use the model seds of @xcite ( see @xcite for a detailed discussion of predicting ir luminosities from _ spitzer _ data ) . our final selection of galaxies consists of those sdss / mpa galaxies with detections in the far - uv ( @xmath15 , @xmath16 ) and near - uv ( @xmath17 , @xmath18 ) through 24 bands . this is a total of 467 of the 645 sdss / mpa galaxies within the @xmath19 deg@xmath20 of sdss/_galex_/_spitzer _ overlap . these galaxies have a stellar mass range of @xmath21 @xmath22 and a sfr range of @xmath23 @xmath22@xmath24 , as determined by @xcite and @xcite from the optical spectra and photometry . to simplify the present analysis we do not consider upper limits or selection effects except to note that we are biased against galaxies with very low @xmath25 or @xmath13 . nevertheless , we recover a significant number of galaxies which appear ` old ' and elliptical but have low levels of uv and ir emission . one of the primary motivations for compiling the sample of galaxies described above was to construct a robust and model independent measure of dust attenuation , the so - called infrared excess ( irx , see @xcite for a discussion of the relation of infrared excess to uv attenuation ) . we adopt the definition @xmath26 where @xmath13 is the 8 - 1000 dust luminosity as determined above and @xmath27 is the luminosity in the @xmath28 band ( @xmath29 ) . a second available dust indicator is the h@xmath30to h@xmath31decrement ( see @xcite and references therein ) , measured from the sdss spectra . this dust measure is only well defined for those galaxies with strong emission lines , and can not be easily compared to global galaxy measures due to spectroscopic aperture effects . we do not consider it in the present study . we take the strength of the 4000 break in galaxies as an indicator of sfh @xcite . this has been measured from the deredshifted sdss spectra by @xcite using the ratio of the flux in two narrow bands ( @xmath32 ) centered at 4050 and 3900 . this narrowband color , d@xmath33 , is less sensitive to reddening by dust than broadband colors . it is not , however , completely insensitive to dust effects @xcite . note that d@xmath33 is only measured within the 3 " sdss aperture , which can cause an overestimate of the integrated d@xmath33 for galaxies with moderate bulge to disk ratios . we do not consider in detail the relation between d@xmath33 and more physical measures of galaxy age or sfh ( e.g. the ratio of current to past averaged star formation rate or the specific star formation rate ) , which is metallicity dependent and requires population synthesis modeling . models of galaxy spectra have suggested that the uv color of galaxies can be decomposed into contributions from dust and sfh , though the effect of dust is the dominant contribution @xcite . in figure [ fig : bigfig ] we show the relation between irx and color for different ranges of d@xmath33 . we find a marginal dependence of the irx vs. @xmath34 relation ( analogous to the irx-@xmath5 relation ) on d@xmath33 . only the very oldest galaxies ( red points ) have systematically redder @xmath34 for the same value of irx , but also have larger errors . for galaxies with d@xmath33@xmath35 there appears to be little dependence of the scatter on galaxy sfh . this is similar to the result of @xcite , who use the @xmath36k color as a measure of sfh ( but see below for the effect of dust on the @xmath36nir color ) . this result can be reconciled with the models of @xcite in several ways . first , some stars contributing to the dust heating ( and therefore the ir emission ) may not be contributing to the uv emission or , conversely , some stars contributing to the uv emission are not contributing to the ir emission ( i.e. a decoupling of ir and uv emission ) . this may occur for various reasons including star - dust geometry ( e.g. @xcite ) and contributions to dust heating from older stellar populations . related to this , d@xmath33 may not be measuring the relevant timescale for changes in the uv spectrum . the sfh diagnostic that is likely to be most relevant to changes in the uv spectral shape is the ratio of current sfr to the sfr averaged over the last 100 myr @xcite , which is not probed well by d@xmath33 . second , weak agn may affect the uv colors . third , the entire range of @xmath34 color is @xmath37 mag , compared to a median error of @xmath38 mag . the scatter due to measurement error may obscure a trend in the scatter with sfh . the results of @xcite were based on a library of 95000 model spectra , which may not be well represented by our sample . these and other possibilities will be investigated in future work . for @xmath39 color it is easy to see the effect of sfh that was predicted by @xcite for the @xmath34 color . at a given low d@xmath33 ( i.e. younger mean stellar age , purple points ) the relation between dust and color is clear , and has low scatter . this relation is closely related to the effective attenuation curve of the dust in these galaxies @xcite . for intermediate d@xmath33 ( i.e. intermediate mean stellar age , orange or green points ) the relation between dust and color persists with low scatter , but the entire relation is shifted to redder color . this is presumably because of the redder intrinsic spectrum of an older stellar population , on top of which the effect of dust attenuation on color persists relatively unchanged . for @xmath40 the ratio of the median error to the range in color is significantly smaller than for @xmath34 , making trends with color easier to see . at the largest d@xmath33 the scatter in the irx - color relation increases this may be due to several causes . first , there are larger errors in irx ( and color ) for these galaxies , which have relatively little star formation and are systematically dimmer in @xmath15 and at 24 . second , the effects of metallicity on d@xmath33 become more pronounced at high d@xmath33 . third , some uv emission may be due to evolved populations ( e.g. bhb stars ) . finally , these galaxies may well host agn that affect the ir and/or uv emission , changing irx . in figure [ fig : model ] we present another projection of the dust - sfh - color relation . here we show d@xmath33 as a function of @xmath39 color for different ranges of irx . irx appears to be well determined using just @xmath39 and d@xmath33 . the relation between d@xmath33 and @xmath40 shifts to redder color for galaxies with more dust , while the slope of the relation remains nearly constant . galaxies in the lower right of this plot are predominantly dusty star - forming galaxies . galaxies in the lower left are typically blue star - forming galaxies or dwarfs / irregulars . galaxies in the upper right of figure [ fig : model ] are red - sequence galaxies composed primarily of old stars , with very low levels of uv and ir emission . we have conducted quantitative parametric fits to the relation between dust attenuation , sfh , and the @xmath41 color . considering a simple effective attenuation law we are motivated to consider polynomial fits to the color ( @xmath42 ) , d@xmath33 , and an expression of irx that is linearly related to color @xmath43 where @xmath44 is the bolometric correction to the uv and the bolometric correction to the ir has been made in [ sec : data ] ( see @xcite for a discussion of the relation between irx and the true uv attenuation @xmath45 ) . we treat the color and d@xmath33 as independent variables ( since the errors are much smaller than for irx ) . we assume either a linear or quadratic form for the relation between color and d@xmath33 . the results are given in table [ table : fit ] . examination of the residuals shows that a cross term ( @xmath46d@xmath33@xmath47 ) is required , and fits including such a term are also given in table [ table : fit ] . the fit from the 3rd row of table [ table : fit ] is overplotted in figure [ fig : model ] , and the residuals in irx as a function of color and d@xmath33 are shown in figure [ fig : fit ] . figure [ fig : fit ] shows that the fit is poorer for redder colors and larger d@xmath33 . this may be due to the effects listed in [ sec : irx_color ] for the @xmath41 color , as well as the effect of aperture on d@xmath33 ( [ sec : sfh ] ) . at bluer colors ( @xmath41@xmath48 ) and smaller d@xmath33 the residuals are lower , @xmath49 rms in @xmath50 . table [ table : fit ] also includes fits using the @xmath51 color since this is more easily measured for a large sample of galex observed sdss galaxies ( e.g. @xcite ) . the behavior is similar to the @xmath41 color though the residuals are larger . cccccccc + 1.31 & -3.46 & & 0.84 & & 0.57 & 0.33 + 1.21 & -1.59 & -2.73 & 0.83 & & 0.55 & 0.31 + 1.08 & -1.80 & & 1.03 & -0.72 & 0.51 & 0.31 + 1.07 & -2.39 & 1.59 & 1.10 & -0.94 & 0.50 & 0.32 + + 1.60 & -3.27 & & 0.87 & & 0.75 & 0.43 + 1.43 & -0.58 & -4.12 & 0.89 & & 0.70 & 0.39 + 1.27 & -1.56 & & 1.35 & -1.24 & 0.65 & 0.38 + 1.25 & -2.91 & 3.30 & 1.56 & -1.82 & 0.64 & 0.39 we have used a simple treatment of attenuation to motivate our fits that does not self consistently consider , e.g. , the heating of dust by old stars . also , d@xmath33 is subject to aperture effects that may affect the fits . it is thus difficult to interpret the fit coefficients as physical parameters . the fits in table [ table : fit ] are referenced to d@xmath33@xmath52 , @xmath53 because this region is typical of ` blue sequence ' galaxies . the constant a term thus gives the typical @xmath50 for such galaxies . the b term ( with contributions from the c and e terms when present ) gives the trend in @xmath50 with d@xmath33 , and is negative since galaxies that are intrinsically redder have less attenuation for a given color . similarly , d gives the trend in @xmath50 with color , and is positive as a redder galaxy at a given d@xmath33 has more attenuation . while we recover only a weak sfh dependence for the short wavelength - baseline uv color , we have shown that long wavelength - baseline colors specifically @xmath39 can be decomposed into contributions from dust and sfh with low scatter . this is the dust - sfh - color relation . such a decomposition is possible due to the use of a relatively dust - insensitive ( though perhaps metallicity dependent ) sfh indicator in combination with a robust measure of the dust attenuation in galaxies . at large d@xmath33 the relation between dust , sfh , and color is more scattered , suggesting that an additional parameter may be necessary to explain the color of such galaxies , or that our sfh and/or dust indicators become less reliable here . deep _ spitzer _ irac data , when combined with optical data , is ideally suited to measuring the @xmath39 color for galaxies at @xmath54 , allowing similar analyses for galaxies at much earlier epochs ( e.g. @xcite and @xcite ) . the mpa / jhu collaboration for sdss studies has very generously made their catalogs publicly available . the publicly available _ spitzer _ data obtained and reduced by the swire team have been essential to this work . we gratefully acknowledge nasa s support for construction , operation , and science analysis for the _ galex _ mission , developed in cooperation with the centre national detudes spatiale of france and the korean ministry of science and technology . bdj was supported by nasa gsrp grant nng05go43h .	we combine data from sdss and the _ galex _ and _ spitzer _ observatories to create a sample of galaxies observed homogeneously from the uv to the far - ir . this sample , consisting of @xmath0460 galaxies observed spectroscopically by sdss provides a multiwavelength ( 0.15 - 24 @xmath1 m ) view of obscured and unobscured star formation in nearby ( @xmath2 ) galaxies with sfrs ranging from 0.01 to 100 m@xmath3 yr@xmath4 . we calculate a robust dust measure from the infrared to uv ratio ( irx ) and explore the influence of star formation history ( sfh ) on the dust - uv color relation ( i.e. the irx-@xmath5 relation ) . we find that the uv colors of galaxies are only weakly dependent on their sfh as measured by the 4000 break . however , we find that the contributions of dust and sfh are distinguishable when colors at widely separated wavelengths ( e.g. 0.23 - 3.6 ) are introduced . we show this explicitly by recasting the irx-@xmath5 relation as a more general irx - sfh - color relation , which we examine in different projections . we also determine simple fits to this relation .
hamiltonian structures play a fundamental role in mathematical physics . it s enough to recall a few examples : classical mechanics , electrodynamics , quantum mechanics , hydrodynamics and general relativity . however , when applying the classical methods and technics of symplectic geometry to pdes , one faces significant difficulties , both analytical and conceptual . part of the problem is that symplectic forms that arise in many applications are weak symplectic forms on infinite dimensional manifolds . more importantly , often integral curves of pdes are not differentiable in time in the function spaces one would normally use ; in the linear case , this corresponds to the fact that the operators involved are unbounded . stock examples include the euler and klein - gordon equations . when dealing with such systems one has to pay careful attention to domains of definitions as many standard formulas become only formal relationships . their justification is often cumbersome and requires some ad hoc methods . the goal of this paper is to contribute to the development of techniques that are useful for the treatment of nonlinear pdes with non - differentiable ( in time ) solutions and build a framework that allows a systematic and rigorous study of such systems and is applicable to the broad range of physical phenomena . previous work in this vein is @xcite . specifically , this article is devoted to the study of the euler equations for an ideal fluid on the compact manifold , the example that provides the main inspiration and motivation . the goal is to understand in what exact sense ( if any ) the flow generated by euler s equation consists of poisson maps . since the classic work of arnold @xcite , it has been known that _ formally _ the euler equation could be viewed as a hamiltonian system . ( expositions of this may be found in @xcite and @xcite ) . the work of @xcite showed the remarkable fact that in appropriate function spaces , the flow of the euler equations in lagrangian representation ( in sobolev function spaces @xmath6 for @xmath7 ) is given by a smooth vector field and hence all the difficulties are resolved in that context . this work also shows that one can perform a reduction ( euler - poincar reduction ) to eulerian representation to rigorously derive that the solutions obtained this way satisfy the euler equations ( taking into account one derivative loss due to the reduction procedure ) . from the work of @xcite , the reduced flow of the euler equations in @xmath6 are known to form a continuous flow in @xmath6 ( both in time and in the initial velocity field ) , and regarded as maps from @xmath6 to @xmath8 , they are @xmath3 . another remarkable property of the solutions also follows from this same work namely that the individual particle trajectories are @xmath9 in time , a fact not so easy to see directly in eulerian representation ( see @xcite ) . while a version of the _ symplectic nature _ of the flow of the euler equations follows directly from the results in @xcite ( taking into account the loss of one derivative ) , it is not so clear that there is a well defined poisson sense for the results . in fact , the work of @xcite ( and many subsequent papers by other authors ) shows that in the poisson context , this derivative loss is a nontrivial issue in defining a good sense in which one has a poisson manifold and in which the euler equations then define a hamiltonian system in the poisson sense . _ the main purposes of this paper is to fill this gap by means of a nonsmooth lie - poisson reduction procedure on appropriate classes of functions . _ this article has the following structure . in [ s : eulersol ] we give important background information on euler equation and manifolds of diffeomorphisms . then , we recall the basic ideas of poisson reduction in [ s : poisson ] . our results are presented in next two sections . in [ s : weakpoisson ] we prove that tangent bundle of a weak riemannian manifold carries a poisson structure in an appropriate sense , provided that the manifold possesses a _ smooth _ riemannian connection . the later requirement is fulfilled on the groups of diffeomorphisms according to the work of @xcite . in [ s : eulerflow ] we utilize this result to show that the flow of euler equation is poisson in an appropriate sense . we conclude with short discussion of presented results in [ s : future ] . in this section we present some classical results concerning the euler equation that motivated our study . the notation and exposition follows @xcite . the euler equations on compact manifold are traditionally formulated in the following way . let @xmath10 be a compact riemannian @xmath2-manifold possibly with boundary @xmath11 . find a time dependent vector field @xmath12 , ( which has an associated flow denoted @xmath13 ) such that 1 . @xmath14 is a given initial condition with @xmath15 2 . the euler equations hold : @xmath16 for some scalar function @xmath17 ( the pressure ) , 3 . @xmath18 , and 4 . @xmath19 is parallel to @xmath11 . it is standard that above equation can be formally rewritten as an ode on the space of divergence free vector fields with a derivative loss . but it was discovered by @xcite that this is literally true with no derivative loss in lagrangian representation . we recall how this proceeds . let @xmath20 be a volume form on the manifold @xmath10 . let @xmath21 denote the space of mappings of sobolev class @xmath22 from an @xmath2-manifold @xmath10 to a manifold @xmath23 . for @xmath24 , let @xmath25 @xmath26 then both @xmath27 are smooth infinite dimensional manifolds and topological groups , moreover @xmath28 is a closed submanifold and a subgroup of @xmath29 . let @xmath30 and @xmath31 be the canonical projections and let @xmath32 be the identity element of the groups @xmath33 . then @xmath34 @xmath35 where @xmath36 denotes the space of @xmath4 divergence free vector fields on @xmath10 that are parallel to the boundary . a given riemannian metric on @xmath10 induces a right invariant weak riemannian metric on @xmath28 given by @xmath37 for @xmath38 where scalar product under the integral sign is taken in @xmath10 . as was shown in @xcite , @xmath28 possesses a smooth riemannian connection and , as a consequence , a smooth spray , which we will denote @xmath39 . [ t : flow ] for @xmath5 , the weak riemannian metric has a @xmath9 spray @xmath40 . let @xmath41 be the ( local , @xmath9 ) flow of @xmath39 . let @xmath42 ( the material velocity field ) and @xmath43 ( the particle position field ) . then the solution of the euler equation with initial condition @xmath44 is given by @xmath45 from the properties of the diffeomorphism group , one sees that this result shows that the euler equations ( [ euler_equations ] ) are well - posed in @xmath6 in eulerian representation . first , recall the following basic and simple result about poisson reduction ( see , for example , @xcite ) . suppose that @xmath46 is a lie group that acts on a poisson manifold @xmath47 and that for each @xmath48 the action map @xmath49 is a poisson map . suppose that the quotient @xmath50 is a smooth manifold and the projection @xmath51 is a submersion . then , there is a unique poisson structure @xmath52 on @xmath50 such that @xmath53 is a poisson map . it is given by @xmath54 where @xmath55 is a poisson bracket in @xmath47 and @xmath56 is a set of smooth functions on @xmath50 . if @xmath57 is a hamiltonian vector field for a g - invariant hamiltonian @xmath58 , then @xmath53 also induces reduction of dynamics . there is a function @xmath59 such that @xmath60 . since @xmath53 is a poisson map it transforms @xmath57 on @xmath47 to @xmath61 on @xmath50 , that is , @xmath62 . denoting the flow of @xmath57 by @xmath63 and the flow of @xmath61 by @xmath64 we obtain commutative diagram @xmath65 our strategy is to apply the above procedure to the context of fluids . to do so , define the map @xmath66 via @xmath67 where @xmath68 . let @xmath69 be given by @xmath70 for @xmath71 by proposition [ t : flow ] , @xmath64 is the flow of euler equation on @xmath72 , i.e. @xmath73 satisfies the euler equations ( [ euler_equations ] ) . it is clear from the preceding developments that @xmath63 ( as a flow of a spray ) is a flow of hamiltonian vector field on @xmath74 . the following commutative diagram @xmath75 suggests that the flow of euler equation itself , which is obtained from @xmath63 via poisson reduction , should be a hamiltonian flow in the sense of poisson manifolds and this is certainly formally true ( see , for instance @xcite for both the case considered here as well as the case of free boundary problems ) . however , as noted in this reference and elsewhere , there are difficulties in finding the right class of functions so that one gets a poisson structure in a precise sense . to justify the formal insight in precise function spaces , one has to overcome two hurdles . the first hurdle is that @xmath76 is only a weak symplectic manifold , and therefore does not necessary carry a poisson bracket in any obvious way without special ad hoc hypotheses such as `` the needed functional derivatives exist '' which have long been recognized as awkward at best . the second hurdle is that @xmath76 is not a lie group in the usual sense ( left multiplication is not smooth ) , and @xmath53 is not a smooth map ( inversion in @xmath28 is not smooth ) . therefore , the well developed theory of poisson and lie - poisson reduction is not directly applicable in this case , even though the loss of derivatives one suffers from these transformations is well understood . the main point of this paper is to resolve these difficulties in what we believe is a satisfactory way . we do this in the following sections . let @xmath77 be a weak riemannian manifold modelled on banach space @xmath78 with metric @xmath79 . then @xmath80 possesses a canonical _ weak _ symplectic form that is given in charts by the following standard formula ( see , e.g. , @xcite ) : @xmath81 where @xmath82 , @xmath83 . for a smooth function @xmath84 on a ( strong ) symplectic manifold @xmath85 , let @xmath86 denote its hamiltonian vector field . then @xmath87 makes @xmath88 into a poisson manifold . since @xmath89 is weak , formula [ f : symppoisson ] does not automatically define poisson bracket @xmath90 for arbitrary functions @xmath91 since @xmath92 may fail to exist and even if they do , one has to make additional hypotheses to obtain the jacobi identity . however , under the two additional hypothesis : 1 . @xmath77 has smooth riemannian connection ; 2 . the inclusion @xmath93 ( the literal dual space ) via @xmath94 is dense , it will be shown that one can define a poisson bracket on the subalgebra @xmath95 of @xmath96 . here @xmath97 are covariant partial derivatives on @xmath80 , the definition of which will be given below . this newly defined bracket _ makes @xmath98 into a lie algebra and retains essential dynamical properties of a `` true '' poisson bracket , including the jacobi identity and the fact that flows of hamiltonian vector fields are poisson maps and , of course , energy is conserved . _ moreover , we will show that the bracket indeed is related to the canonical weak symplectic form in the way that one would expect . in the following we assume that conditions ( 1 ) and ( 2 ) are satisfied . [ [ covariant - partial - derivatives . ] ] covariant partial derivatives . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + first , we introduce covariant partial derivatives on @xmath80 . let @xmath99 and @xmath100 be natural projections , @xmath101 be a christoffel map and @xmath102 be a connector map . in local representation , @xmath103 define @xmath104 by @xmath105 it is standard that @xmath106 is a diffeomorphism ( see @xcite ) . for @xmath107 we set @xmath108 @xmath109 in local representation , this reads @xmath110 and @xmath111 similarly , for @xmath112 we define @xmath113 ( here @xmath114 is the space of linear maps @xmath115 ) by @xmath116 @xmath117 the following lemmas are readily verified . [ t : prop1 ] let @xmath118 be a vector field on @xmath80 , @xmath119 be a vector field on @xmath120 , @xmath121 . then @xmath122 [ t : prop2 ] for @xmath123 , we have @xmath124 where @xmath125 is the parallel translation of @xmath126 along the curve @xmath13 with @xmath127 . let @xmath128 now we can define the bracket @xmath129 via @xmath130 [ [ preliminaries - on - the - poisson - structure . ] ] preliminaries on the poisson structure . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the following is the first main result . [ t : prop4 ] the bracket maps @xmath131 into @xmath132 and also maps @xmath133 into @xmath134 . [ [ remark . ] ] remark . + + + + + + + by definition of the covariant partial derivatives , @xmath135 for @xmath136 . the theorem asserts that if @xmath137 then , in fact , @xmath138 , i.e. there are @xmath139 such that @xmath140 and the maps @xmath141 have appropriate smoothness . define operator @xmath142 . this definition extends the usual notion of covariant derivative from vector fields along curves on @xmath77 to arbitrary curves on @xmath80 . let @xmath143 and @xmath137 . choosing @xmath125 as in lemma [ t : prop2 ] , we obtain @xmath144 to proceed further , we need to calculate the quantity @xmath145 where @xmath146 is an arbitrary element of @xmath147 . let @xmath148 be a parametric surface in @xmath80 with the following properties : 1 . @xmath149 ; 2 . @xmath150 is a parallel translation of @xmath126 ; 3 . @xmath151 is a parallel translation of @xmath152 ; 4 . @xmath153 for all @xmath22 ; 5 . @xmath148 is a parallel translation of @xmath150 for all @xmath22 . then , keeping in mind lemmas [ t : prop1 ] , [ t : prop2 ] and symmetry of riemannian connection , one checks the following : @xmath154 @xmath155 @xmath156\ ] ] @xmath157\ ] ] @xmath158 @xmath159 [ t : lemma1,2 ] ) . let @xmath160 denote the ricci curvature tensor . then @xmath161 @xmath162 by construction of @xmath148 , we have @xmath163 . applying lemma [ t : lemma1,2 ] we obtain @xmath164 @xmath165 thus , @xmath166 by bianchi s identity . similar calculations yield @xmath167 @xmath168 substituting this into the formulas for @xmath169 and using bianchi s identity once again , we get @xmath170 similarly , @xmath171 @xmath172 as @xmath173 is smooth , the statement of the theorem follows . [ [ hamiltonian - vector - fields . ] ] hamiltonian vector fields . + + + + + + + + + + + + + + + + + + + + + + + + + + the smoothness structure of hamiltonian vector fields is given as follows . [ t : prop5,6 ] the vector field @xmath57 is a @xmath174 hamiltonian vector field ( with respect to canonical weak symplectic form ) on @xmath80 of class @xmath174 if and only if @xmath175 . moreover , @xmath176 in local representation , we have @xmath177 indeed , @xmath178 substituting this expression into the formula for @xmath89 and using the symmetry of @xmath179 we obtain the desired result . let @xmath180 be a hamiltonian vector field , @xmath181 be arbitrary . then @xmath182 on the other hand , by lemma [ t : prop1 ] @xmath183 setting @xmath184 and comparing the above expressions we see that @xmath185 similarly , setting @xmath186 yields @xmath187 thus , @xmath188 . conversely , let @xmath189 . defining a vector field @xmath57 by formula [ e : ham ] and substituting into formula [ e : omega ] one obtains for arbitrary vector @xmath190 @xmath191 [ t : prop7 ] let @xmath192 be arbitrary . then @xmath193 by proposition [ t : prop5,6 ] , the vector fields @xmath92 are defined whenever @xmath194 is . then @xmath195 [ t : pbrack ] the bracket @xmath196 is antisymmetric , bilinear , derivation on each factor and makes @xmath134 into a lie - algebra . antisymmetry , linearity and property of being derivation follows directly from the definition of the bracket . by theorem [ t : prop4 ] @xmath196 leaves @xmath134 invariant . then , jacobi identity follows from proposition [ t : prop7 ] in the usual way , for example as in @xcite . now , @xmath80 has both symplectic and poisson structures , and therefore two generally different definitions of hamiltonian vector fields . we need to check that in our case these coincide . to do so , let @xmath197 temporarily denote the hamiltonian vector field with respect to poisson structure @xmath196 and @xmath86 denotes the hamiltonian vector field with respect to canonical symplectic form corresponding to function the @xmath198 . recall , that @xmath197 is defined as a vector field such that @xmath199= { \left \{h , f \right \ } } \quad \forall h\in \mathcal{k}.\ ] ] thus , for all @xmath200 , @xmath201 & = { \frac{\partial h}{\partial \eta } } \cdot \frac{\partial f}{\partial v } - { \frac{\partial h}{\partial v } } \cdot \frac{\partial f}{\partial \eta}\\ & = dh \cdot \mathbf{x}_{f}^p = { \frac{\partial h}{\partial \eta } } \cdot t \tau \mathbf{x}_{f}^p+{\frac{\partial h}{\partial v } } \cdot k \mathbf{x}_{f}^p\end{aligned}\ ] ] and therefore , @xmath202 and @xmath203 . comparing this with formula [ e : ham ] , we see that @xmath204 . finally , from the coordinate expression , it is easy to see that @xmath86 is a well defined @xmath174 vector field for any @xmath205 . previously we established that classes @xmath206 are preserved under bracketing . unfortunately , for @xmath205 and a diffeomorphism @xmath207 the composition @xmath208 does not have to be in any class @xmath209 . one can , however , compose with _ symplectic diffeomorphisms . _ [ t : compdif ] let @xmath210 be a symplectic @xmath174 diffeomorphism , @xmath205 . then @xmath211 . we have @xmath212 and so by proposition [ t : prop5,6 ] , @xmath213 . [ t : flowpoisson ] let @xmath63 be a flow of a smooth hamiltonian vector field on @xmath214 . then @xmath63 is a poisson , i.e. for all @xmath215 @xmath216 @xmath63 is symplectic with respect to the weak riemannian form . since @xmath63 preserves class @xmath134 , the statement follows from jacobi identity by the usual argument . as we stated earlier , in @xcite it is shown that @xmath28 carries a smooth riemannian connection , and therefore the results of the previous section apply . therefore , by those results , the space @xmath74 carries a poisson structure ( in the precise sense given there ) which we denote @xmath196 . let @xmath217 , @xmath218 , @xmath219 stand for the corresponding connector maps on the underlying manifold @xmath10 , on @xmath29 and @xmath28 respectively , while @xmath220 , @xmath221 , @xmath222 are the corresponding connections and @xmath179 , @xmath223 , @xmath224 are the corresponding christoffel maps . in the following @xmath79 denotes the riemannian metric on @xmath10 , @xmath29 , @xmath28 and an induced scalar product on @xmath225 depending on the context . the relationship between these metrics is given by [ e : scal ] . recall the notation from [ s : poisson ] . namely , let @xmath63 be the flow of the spray on @xmath76 , @xmath64 denote the flow of euler equation on @xmath72 and @xmath226 , @xmath227 . recall also that we have the commutative diagram [ t : commdiag ] the following diagram is commutative : @xmath228 now we prepare and recall from @xcite some useful lemmas . let @xmath229 . define @xmath230 via @xmath231 . then @xmath232 indeed , notice that @xmath233 by right invariance of the spray . thus , @xmath234 is an integral curve of @xmath39 . since @xmath235 , the statement of the lemma follows from uniqueness of integral curves . recall that by definition , @xmath236 for all @xmath237 . let @xmath238 . then , using the preceding lemma , we obtain @xmath239 notice , that @xmath240 for any @xmath229 . indeed , @xmath241 thus @xmath242 and the proposition is proved . [ [ a - poisson - structure - on - the - lie - algebra . ] ] a poisson structure on the lie algebra . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + now , we construct a poisson bracket @xmath243 on @xmath72 so that @xmath53 is a poisson map . for @xmath244 such that @xmath245 define @xmath246 as in [ s : weakpoisson ] , define @xmath247 and @xmath248 let @xmath249 . then @xmath243 is a bilinear map @xmath250 and a derivation on each factor . moreover , it satisfies jacobi identity on @xmath251 , that is for all @xmath252 , and @xmath253 , @xmath254 let @xmath255 . recall , that for @xmath256 , @xmath257 is an algebra . thus , @xmath258 is a bilinear bounded map @xmath259 ( and @xmath260 ) , hence smooth . this implies that @xmath261 bilinearity and derivation property of @xmath243 trivially follows from properties of @xmath262 and @xmath79 . now we calculate @xmath263 . let @xmath264 . since @xmath265 , the frchet derivative of @xmath266 exists and coincides with its gateaux derivative . thus , by bilinearity of scalar product and @xmath220 , @xmath267 [ t : lem5 ] let @xmath268 , @xmath249 , and let @xmath269 be @xmath4 vector fields on m. then @xmath270 by the sobolev theorems , @xmath118 is a @xmath271 vector field on @xmath10 . by properties of the riemannian connection , for all @xmath272 @xmath273 thus , @xmath274 let @xmath275 be a flow of @xmath118 on @xmath10 . since @xmath118 is divergence free , @xmath20 is @xmath275 invariant , i.e. @xmath276 , where @xmath277 denotes a pullback by @xmath275 . then @xmath278 [ t : lem6 ] let @xmath279 , @xmath280 . then for all @xmath281 @xmath282 we compute as follows : @xmath283 [ t : hodge ] . let @xmath118 be an @xmath4 vector field on @xmath10 , @xmath284 . there is an @xmath285 function @xmath286 and an @xmath4 vector field @xmath119 with @xmath119 divergence free , such that @xmath287 further , the projection maps @xmath288 @xmath289 are continuous linear maps on @xmath290 . the decomposition is orthogonal in @xmath291 sense , that is for all @xmath292 @xmath293 [ t : lem7 ] there is a bilinear continuous map @xmath294 @xmath295 such that for all @xmath296 @xmath297 fix coordinate system @xmath298 on @xmath10 and let @xmath299 denote components of metric tensor , @xmath300 denote components of vector field @xmath301 in the chosen system . let @xmath302 ( as usually , the summation on repeated indexes is understood ) . then @xmath303 where @xmath304 since @xmath4 is an algebra for @xmath305 it follows that @xmath306 is an @xmath4 vector field . now we set @xmath307 and use [ e : hodge ] . by lemmata [ t : lem5]-[t : lem7 ] , we have @xmath308 thus for any @xmath309 , @xmath310 \\ & \quad + ddf(v)\cdot b(dg(v),v)-ddg(v)\cdot b(df(v),v ) \\ & \quad + ddg(v ) \cdot p_e { \nabla}_{df(v)}v - ddf(v ) \cdot p_e { \nabla}_{dg(v)}v,\end{aligned}\ ] ] and hence @xmath311 and @xmath312 . [ [ remark.-1 ] ] remark . + + + + + + + if @xmath313 , then @xmath314 , @xmath315 are @xmath174 as maps @xmath316 , hence @xmath317 . now we prove the jacobi identity . to simplify notation , we set @xmath318 moreover , since in the following argument all functions are evaluated at the same point @xmath319 , we will write @xmath320 instead of @xmath321 , etc . by lemmata [ t : lem6]-[t : lem7 ] , we obtain @xmath322,{\nabla}_{f}-b_f \right \rangle}+ { \left \langleddg \cdot ( b_h - { \nabla}_{h } ) , { \nabla}_{f}-b_f \right \rangle } \\ & \qquad + { \left \langleddh \cdot ( { \nabla}_{g } - b_g),{\nabla}_{f}-b_f \right \rangle } \\ & = { \left \langle[dh , dg],{\nabla}_{f}-b_f \right \rangle}+d_{ghf}-d_{hfg},\end{aligned}\ ] ] where @xmath323 and @xmath324 $ ] is a lie bracket of vector fields on @xmath10 . notice that lie bracket of divergence free vector fields is divergence free . for @xmath325 @xmath326,{\nabla}_{f}-b_f \right \rangle}={\left \langle\left[\left[dh(v),dg(v)\right],df(v)\right],v \right \rangle}.\ ] ] since terms of type @xmath327 cancel out in the jacobi cycle @xmath328 and so the jacobi identity for bracket @xmath243 follows from the jacobi identity for vector fields . however , for @xmath329 lie bracket of @xmath330 and @xmath331 is an @xmath332 vector field , hence merely continuous and therefore @xmath333,df(v)\right]$ ] may fail to exist . therefore , in this case more care is needed . let @xmath334 @xmath335}v \right \rangle}\ ] ] with this notation in mind , by lemma [ t : lem5 ] and the hodge decomposition @xmath326,{\nabla}_{f } \right \rangle}={\left \langle{\nabla}_{dh}dg-{\nabla}_{dg}dh,{\nabla}_{df}v \right \rangle}= - a_{ghf}+a_{hgf}.\ ] ] similarly , by definition of @xmath336 @xmath326,b_f \right \rangle}= c_{fhg}.\ ] ] by a well known formula for riemannian connection , @xmath337}z,\ ] ] for all sufficiently smooth vector fields @xmath338 . thus , @xmath339}v \right \rangle}= c_{fgh}.\ ] ] thus , @xmath340 and so @xmath341 [ [ remark.-2 ] ] remark . + + + + + + + if @xmath342 , @xmath249 , then by lemma [ t : lem5 ] @xmath343,v \right \rangle}.\ ] ] this shows that bracket @xmath243 is naturally related to lie - poisson bracket on @xmath344 . now we establish the relationship between poisson bracket @xmath243 on @xmath72 that we just introduced and poisson bracket @xmath196 on @xmath28 . for @xmath345 define @xmath346 [ t : eqdef ] define the function spaces @xmath347 and @xmath348 then @xmath349 for @xmath350 @xmath351 and for all @xmath352 @xmath353 without loss of generality @xmath354 . since @xmath53 is not even a @xmath271 function @xmath355 it is not obvious that @xmath356 is defined . however , differentiating @xmath357 and @xmath358 as functions @xmath359 one obtains the required result . [ t : dfrv ] under the assumptions of the theorem , @xmath360 it is well known ( @xcite ) that @xmath361 . notice , that for @xmath362 , @xmath363 where time derivative is taken in @xmath76 . by lemma [ t : prop1 ] @xmath364 thus , by right invariance of the metric on @xmath28 @xmath365 [ t : dfrn ] under the assumptions of the theorem @xmath366 \right \rangle}_e\ ] ] that is , @xmath367 first , we calculate @xmath368 . let @xmath369 , @xmath370 be a parallel translation of @xmath126 with @xmath371 . recall that @xmath372 then , by lemma [ t : prop2 ] , @xmath373 since connection on @xmath28 is right invariant , i.e. , @xmath374 we have @xmath375=\left [ \widetilde{k } \frac{d}{dt}_{{t=0 } } v_t \right ] \circ \eta^{-1 } = 0.\ ] ] by lemma [ t : prop1 ] @xmath376 combining above equalities together , we get @xmath377.\ ] ] @xmath378 \right \rangle}_e.\ ] ] we claim that for all @xmath379 @xmath380 \right \rangle}={\left \langlez,{\nabla}_{y}x \right \rangle}.\ ] ] recall that by construction ( see @xcite ) , @xmath381 @xmath382 @xmath383 by a well known formula of differential geometry , we have @xmath384 and hence @xmath385 = p_e[{\nabla}_{y}x].\ ] ] by the hodge decomposition @xmath386 \right \rangle } = { \left \langlez,{\nabla}_{y}x \right \rangle}={\left \langleb(z , x),y \right \rangle}.\ ] ] by the above developments and right invariance of metric on @xmath28 , we have @xmath387 calculating @xmath356 at @xmath309 by lemmata [ t : dfrv],[t : dfrn ] , we obtain @xmath388 [ t : pipoisson ] map @xmath389 is a poisson map , i.e. for all @xmath390 pointwise in @xmath391 @xmath392 @xmath393 since @xmath53 is the identity on @xmath72 , the statement follows immediately from theorem [ t : eqdef ] . [ t : fpoisson ] let @xmath394 and @xmath395 are such that @xmath396 , @xmath397 , @xmath398 , @xmath399 , @xmath400 . then @xmath401 in particular , @xmath63 preserves @xmath402 and for @xmath403 pointwise in @xmath391 @xmath404 without loss of generality @xmath405 . first , we notice that covariant partial derivatives of @xmath406 at @xmath407 are elements of @xmath76 . indeed , @xmath408 there is a function @xmath409 such that @xmath410 thus , @xmath411 however , by proposition [ t : compdif ] @xmath412 for any @xmath413 , hence there is @xmath414 such that for all @xmath19 , @xmath415 in a similar sense , one shows that @xmath416 . thus , @xmath417 is well defined and depends only on values of @xmath418 calculated at point @xmath419 . however , @xmath420 also depends only on values of covariant partial derivatives at @xmath419 . then , we choose @xmath421 such that @xmath422 the equality @xmath423 follows from proposition [ t : flowpoisson ] . by the preceding arguments , the same holds if we replace @xmath424 with @xmath425 . this concludes the first part of the proposition . the second part then follows . [ s : redpoisson ] the map @xmath426 is poisson with respect to the bracket @xmath243 . . then @xmath428 . by proposition [ t : fpoisson ] @xmath429 and we have pointwise in @xmath430 : @xmath431 in the previous sections we successfully implemented a nonsmooth lie - poisson reduction technique for the study of the euler equations of ideal fluid flow . this enabled us to find a precise sense in which the flow of euler equation on the lie algebra of divergence free vector fields ( parallel to the boundary of the fluid region ) is a hamiltonian system in the poisson sense and that the flow consists of poisson maps , despite the fact that this flow is believed ( as maps from @xmath4 to @xmath4 ) to be continuous , but not differentiable . a key part of this process was to introduce a poisson structure on the space of divergence free vector fields . as one would expect from the bracket derived via a type of lie - poisson reduction , this bracket is closely related to the formal lie - poisson bracket on the dual to the lie algebra of divergence free vector fields . even though we consider only euler s equation , the technique developed here is directly applicable to several other important systems those which can be written as an ode on groups of diffeomorphisms , such as the following : 1 . the camassa - holm ( ch ) equation on @xmath432see @xcite : @xmath433 2 . the averaged euler equations ( or the lae-@xmath434 equations)see @xcite : @xmath435 where @xmath436 and @xmath12 satisfies appropriate boundary conditions , such as the no - slip conditions @xmath437 on @xmath438 . 3 . the epdiff equation ( also called the averaged template matching equation ) on a compact manifold @xmath10see @xcite and @xcite : @xmath439 with appropriate boundary conditions , such as the no - slip conditions @xmath437 on @xmath11 . the epdiff equations reduce to the ch equations in the case @xmath440 . these equations may be derived as the right reduction to the identity of the geodesic motion on the appropriate lie group ( see , for example , @xcite and @xcite for the case of the ch equations ) , and the preceding references for the other equations . the crucial technical fact that enables our methods to work in both cases is the smoothness of the spray on the lie group . for the case of the ch equations and the lae-@xmath434 equations on regions with no boundary , this is due to @xcite and for regions with boundary to @xcite . for the case of the epdiff equations , a rather convincing plausibility argument is given @xcite . one important direction in which we would like to pursue these ideas is that of nonsmooth solutions . even for the ideal euler equations , this is interesting because of the singular solutions , such as point vortices , vortex filaments and sheets . they clearly have themselves an interesting poisson structure , as was investigated by @xcite and @xcite . there are similar interesting singular solutions for the epdiff equations , whose geometry is investigated in @xcite . it would be very interesting if , on the smaller spaces appropriate for these classes of singular solutions that are introduced in these references , the smooth spray property still holds and , if that is the case , whether or not one could then carry out the program in the present paper . another interesting direction for the present research is to the case of free boundary problems , a notoriously difficult case for infinite dimensional poisson structures , even at the formal level ( see @xcite , @xcite , @xcite and @xcite . ) holm , d.d . and j. e. marsden [ 2003 ] , momentum maps and measure valued solutions ( peakons , filaments , and sheets ) of the euler - poincar equations for the diffeomorphism group . in marsden , j. e. and t. s. ratiu , editors , _ festshrift for alan weinstein ( to appear)_. birkhuser boston .	this paper provides a precise sense in which the time @xmath0 map for the euler equations of an ideal fluid in a region in @xmath1 ( or a smooth compact @xmath2-manifold with boundary ) is a poisson map relative to the lie - poisson bracket associated with the group of volume preserving diffeomorphism group . this is interesting and nontrivial because in eulerian representation , the time @xmath0 maps need not be @xmath3 from the sobolev class @xmath4 to itself ( where @xmath5 ) . the idea of how this difficulty is overcome is to exploit the fact that one does have smoothness in the lagrangian representation and then carefully perform a lie - poisson reduction procedure .
nature has always been an inspiration for researchers . in the past , many new nature - inspired algorithms have been developed to solve hard problems in optimization . in general , there are two main concepts developed in bio - inspired computation : 1 . evolutionary algorithms , 2 . swarm intelligence algorithms . the evolutionary algorithms are optimization techniques @xcite that base on the darwin s principle of survival of the fittest @xcite . it states that in nature the fittest indivi- + duals have greater chances to survive . the evolutionary algorithms consist of the following disciplines : genetic algorithms , evolution strategies , genetic programming , evolutionary programming , differential evolution . although all these algorithms or methods have been developed independently , they share similar characteristics ( like variation operators , selection operators ) when solving problems . in fact , the evolutionary algorithms are distinguished by their representation of solutions . for example , the genetic algorithms @xcite support the binary representation of solutions , evolution strategies @xcite and differential evolution @xcite work on real - valued solutions , genetic programming @xcite acts on programs in lisp , while the evolutionary programming @xcite behaves with the finish state automata . the evolutionary algorithms have been applied to a wide range of areas of optimization , modeling and simulation . essentially , the differential evolution has successfully + been employed in the following areas of optimization : function optimization @xcite , large - scale global optimization @xcite , graph coloring @xcite , chemical - process optimization @xcite . swarm intelligence is a collective behaviour of decentralized , self - organized systems , either natural or artificial . swarm intelligence was introduced by beni in 1989 . a lot of algorithms have been proposed since then . the swarm - intelligence algorithms have been applied on continuous as well as combinatorial optimization problems @xcite . the most well - known classes of the swarm - intelligence algorithms are : particle swarm optimization , ant colony optimization , artificial bee colony , firefly algorithm , cuckoo search and bat algorithm . particle swarm optimization has been successfully applied in problems of the antenna design @xcite and electromagnetics @xcite . the ant colony algorithms have also been used in many areas of optimization @xcite @xcite @xcite . the artificial bee colony shows good performance in numerical optimization @xcite @xcite , in large - scale global optimization @xcite and also in combinatorial optimization @xcite @xcite @xcite . the cuckoo search algorithm is a very strong method for function optimization and also for engineering optimization problems @xcite @xcite . the firefly algorithm shows promising results in function optimization and provides good results also in combinatorial optimization @xcite . echolocation is an important feature of bat behaviour . that means that bats emit a sound pulse and listen to the echo bouncing back from obstacles whilst flying . this phenomenon has inspired yang @xcite to develop the bat algorithm ( ba ) . the algorithm shows good results when dealing with lower - dimensional optimization problems , but may become problematic for higher - dimensional problems because of its tending to converge very fast initially . on the other hand , the differential evolution @xcite is a typical evolutionary algorithm with differential mutation , crossover and selection that is being successfully applied to continuous function optimization . to improve the bat - algorithm behaviour for higher - dimensional problems , we hybridized the original bat algorithm by using differential - evolution strategies . this hybridized bat algorithm ( hba ) was tested on a standard set of benchmark functions taken from literature . results of our numerical experiments show that the proposed hba can significantly improve the performance of the original ba , which can be very useful for the future . the structure of the paper is as follows . in section 2 , we introduced the original ba and the differential evolution algorithm and explain some biological foundations of the bat behaviour . in section 3 we describe our novel approach to hybridizing the bat algorithm with differential evolution strategies . in section 4 we illustrate our experiments and discuss the obtained results . the paper ends by presenting our plans for future work on hba development . the bat algorithm was developed by xin - she yang in 2010 @xcite . the algorithm exploits the so - called echolocation of the bats . the bats use sonar echoes to detect and avoid obstacles . it is generally known that sound pulses are transformed into a frequency which reflects from obstacles . the bats navigate by using the time delay from emission to reflection . they typically emit short , loud sound impulses . the pulse rate is usually defined as 10 to 20 times per second . after hitting and reflecting , the bats transform their own pulse into useful information to gauge how far away the prey is . the bats are using wavelengths that vary in the range from 0.7 to 17 mm or inbound frequencies of 20 - 500 khz . to implement the algorithm , the pulse frequency and the rate have to be defined . the pulse rate can be simply determined in the range from 0 to 1 , where 0 means that there is no emission and 1 means that the bats emitting is their maximum @xcite . the bat behaviour can be used to formulate a new ba . yang @xcite used three generalized rules when implementing the bat algorithms : 1 . all the bats use an echolocation to sense the distance and they also guess the difference between the food / prey and background barriers in a somewhat magical way . when searching for their prey , the bats fly randomly with velocity @xmath0 at position @xmath1 with fixed frequency @xmath2 , varying wavelength @xmath3 and loudness @xmath4 . they can automatically adjust the wavelength ( or frequency ) of their emitted pulses and adjust the rate of pulse emission @xmath5 $ ] , depending on the proximity of their target . although the loudness can vary in many ways , we assume that it varies from a large ( positive ) @xmath4 to a minimum constant value @xmath6 . objective function @xmath7 , @xmath8 initialize the bat population @xmath9 and @xmath10 for @xmath11 define pulse frequency @xmath12 $ ] initialize pulse rates @xmath13 and the loudness @xmath14 while ( @xmath15 ) // number of iterations generate new solutions by adjusting frequency and update velocities and locations / solutions [ eq.(2 ) to ( 4 ) ] if(@xmath16 ) select a solution among the best solutions generate a local solution around the best solution end if generate a new solution by flying randomly if(@xmath17 ) accept the new solutions increase @xmath13 and reduce @xmath14 end if rank the bats and find the current best end postprocess results and visualization the original ba is illustrated in algorithm [ bat ] . in this algorithm , the bat behaviour is captured into the fitness function of the problem to be solved . it consists of the following components : * initialization ( lines 2 - 4 ) , * generation of new solutions ( lines 6 - 7 ) , * local search ( lines 8 - 11 ) , * generation of a new solution by flying randomly ( lines 12 - 16 ) and * find the current best solution . initialization of the bat population is performed randomly . generating new solutions is performed by moving virtual bats according to the following equations : @xmath18 where @xmath19 is a uniform distribution . a random walk with direct exploitation is used for the local search that modifies the current best solution according to equation : @xmath20 where @xmath21 is the scaling factor , and @xmath22 the loudness . the local search is launched with the proximity depending on pulse rate @xmath13 . the term in line 13 is similar to the simulated annealing behavior , where the new solution is accepted with some proximity depending on parameter @xmath14 . in line with this , the rate of pulse emission @xmath13 increases and the loudness @xmath14 decreases . both characteristics imitate the natural bats , where the rate of the pulse emission increases and the loudness decreases when a bat finds its prey . mathematically , these characteristics are captured with the following equations : @xmath23,\ ] ] where @xmath24 and @xmath25 are constants . actually , @xmath24 + parameter plays a similar role as the cooling factor in the simulated annealing algorithm that controls the convergence rate of this algorithm . differential evolution ( de)@xcite@xcite is a technique for the optimization introduced by storn and price in 1995 . de optimizes a problem by maintaining a population of candidate solutions and creates new candidate solutions by combining the existing ones according to its simple formulae , and then keeping whichever candidate solution has the best score or fitness on the optimization problem at hand . de supports a differential mutation , a differential crossover and a differential selection . in particular , the differential mutation randomly selects two solutions and adds a scaled difference between these to the third solution . this mutation can be expressed as follows : @xmath26 where @xmath27 $ ] denotes the scaling factor as a positive real number that scales the rate of modification while @xmath28 are randomly selected vectors in the interval @xmath29 . a uniform crossover is employed as a differential crossover by de . the trial vector is built out of parameter values copied from two different solutions . mathematically , this crossover can be expressed as follows : @xmath30 where @xmath31 $ ] controls the fraction of parameters that are copied to the trial solution . note , the relation @xmath32 assures that the trial vector is different from the original solution @xmath33 . mathematically , differential selection can be expressed as follows : @xmath34 in the technical sense , the crossover and mutation can be performed in many ways in de . therefore , a specific notation was used to describe the variety of these methods ( also strategies ) in general . for example , `` de / rand/1/bin '' denotes that the base vector is randomly selected , one vector difference is added to it , and the number of modified parameters in the mutation vector follows the binomial distribution . as mentioned above , in this article we propose a new ba , called hybrid bat algorithm ( hba ) . it was obtained by hybridizing the original ba using the de strategies . the hba pseudo - code is illustrated in algorithm [ hbat ] . objective function @xmath7 , @xmath8 initialize the bat population @xmath9 and @xmath10 for @xmath11 define pulse frequency @xmath12 $ ] initialize pulse rates @xmath13 and the loudness @xmath14 while ( @xmath15 ) // number of iterations generate new solutions by adjusting frequency and updating velocities and locations / solutions [ eq.(2 ) to ( 4 ) ] if(@xmath16 ) * modify the solution using `` de / rand/1/bin '' * end if generate a new solution by flying randomly if(@xmath17 ) accept the new solutions increase @xmath13 and reduce @xmath14 end if rank the bats and find the current best end postprocess results and visualization as a result , hba differs from the original ba in line 9 , where solution is modified using `` de / rand/1/bin '' strategy . the goal of our experiments was to show that hba significantly improves the results of the original ba . for this purpose , two bas were implemented according to specifications given in algorithms [ bat ] and [ hbat ] so that a well - selected set of test functions in the literature are used for optimization benchmarks . the parameters of the two bas were the same . the dimension of the problem significantly affects the results + optimization . to test the impact of the problem dimension on the results , three different sets of dimensions were taken into account , i.e. , @xmath35 , @xmath36 , and @xmath37 . the functions with dimension @xmath35 were limited to maximally 1,000 generations , the functions with dimension @xmath36 to twice as much , while the functions with dimension @xmath37 to 3,000 . the initial loudness was set to @xmath38 and so was also the initial pulse rate ( @xmath39 ) . the frequency was taken from interval @xmath40 $ ] . the algorithms optimized each function 25 times and results were measured according to the best , worst , mean , and medium values in these runs . the standard deviation of the mean values was calculated as well . our test suite consists of five standard functions taken from literature @xcite . the functions in this test suite are represented below . the aim of the function is overcoming failures to optimize each variable independently . this function is multimodal , since the number of local optima increases with the dimensionality . after dimensionalities are sufficiently high ( @xmath41 ) , multimodality seems to disappear and the problem turns to be unimodal . @xmath42 where @xmath43 . the global minimum of the function is at 0 . as in case of the rastrigin s function , the rosenbrock s function too , has its value 0 at the global minimum . the global optimum is inside the parabolic , narrow - shaped flat valley . variables are strongly dependent on each other , since it is difficult to converge the global optimum . @xmath44 where @xmath45 . @xmath46 where @xmath45 . based of the sphere s function , the rastrigin s function adds cosine modulation to create many local minima . because of this feature , the function is multimodal . its global minimum is at 0 . @xmath47 where @xmath45 . the complexity of this function is moderate because of the exponential term that covers its surface with numerous local minima . it is based on the gradient slope . only the algorithm that uses the gradient steepest descent will be trapped in a local optimum . the search strategy , analyzing a wider area , will be able to cross the valley among the optima and achieve better results . @xmath48,\\ \end{aligned } \normalsize\ ] ] where @xmath49 . the global minimum of this function is at 0 . configuration of pc used in our experiments is as follows : * hp pavilion g4 , * processor intel(r ) core(tm ) i5 @ 2.40 ghz , * memory 8 gb , * implemented in c++ . [ cols="^,^,<,<,<,<,<,<",options="header " , ] [ tab : detail ] the results of our extensive numerical experiments are summarized in table [ tab : detail ] . the table represents results of ba and hba algorithms ( column 1 ) solving the test suite of five functions ( denoted as @xmath50 ) with dimensions @xmath51 @xmath52 @xmath53 @xmath54 , respectively , according to the best , worst , mean , median , and standard deviation values . the results of the hba algorithm show that this algorithm significantly improves the results of the original ba algorithm according to almost all the measures safe for the standard deviation in some cases ( e.g. , when using ackley s function ) . there was no statistical analysis of the results made because they are evidently better for hba than for ba . to show how the results of the two algorithms ( i.e. , ba and hba ) vary with the dimensions of the functions , the mean value of functions @xmath55 and @xmath56 with dimensions @xmath35 , @xmath36 , and @xmath37 are presented in figs . [ pic : g1]-[pic : g3 ] . the logarithmic scale is used to display the mean value on the @xmath57-axis . the higher the mean value , the more difficult the function is to solve . with various dimensions . ] as seen from fig . [ pic : g1 ] , the best results are obtained by optimizing function @xmath55 with the dimension @xmath36 , while the worst by optimizing the same function with the highest dimension @xmath37 . with various dimensions . ] it is interesting to note that the mean value of function @xmath58 with dimension @xmath35 is most difficult to obtain for the hba algorithm , while the same function but with dimension @xmath36 is the easiest to solve . oppositely , the results of the original ba algorithm show that by increasing the dimensions , the results become worsen . with various dimensions . ] as seen from fig . [ pic : g2 ] , solving function @xmath56 becomes more efficient when the dimensionality of the problem increases . as a result , the most difficult function to solve is function @xmath56 with dimension @xmath37 . comparing the results obtained with the ba and the hba algorithm shows that the results of hba significantly outperform the results of the original ba algorithm by optimizing the functions @xmath55 , @xmath56 , and @xmath58 . functions @xmath59 and @xmath60 of the hba algorithm are also better , but the difference is not significant . in this paper we improve the bat algorithm , ba , by developing its new variant , the so - called hybrid bat algorithm , hba . hba is a hybrid of ba with de strategies . as shown with our experiments , hba improves significantly the original variant of ba . in future , hba will be tested on a large - scale global optimization . our testing will be extended by using more diverse test function sets and by deepening our parametric study . j. brest , a. zamuda , i. fister , and m.s . maucec . large scale global optimization using self - adaptive differential evolution algorithm . in _ evolutionary computation ( cec ) , 2010 ieee congress on _ , pages 18 . ieee , 2010 . i. fister , i. fister jr , and j.b.v . memetic artificial bee colony algorithm for large - scale global optimization . in _ evolutionary computation ( cec ) , 2012 ieee congress on _ , pages 18 . ieee , 2012 . tasgetiren , q.k . pan , p.n . suganthan , and a.h.l . a discrete artificial bee colony algorithm for the permutation flow shop scheduling problem with total flowtime criterion . in _ evolutionary computation ( cec ) , 2010 ieee congress on _ , pages 18 . ieee , 2010 .	swarm intelligence is a very powerful technique to be used for optimization purposes . in this paper we present a new swarm intelligence algorithm , based on the bat algorithm . the bat algorithm is hybridized with differential evolution strategies . besides showing very promising results of the standard benchmark functions , this hybridization also significantly improves the original bat algorithm . _ to cite paper as follows : i. jr . fister , d. fister , x .- s yang . a hybrid bat algorithm . elektrotehniki vestnik , 2013 , in press .
it is known that the solution of the kdv equation @xmath0 has a weak limit as @xmath1 while the initial values @xmath2 are fixed . this weak limit is described by hyperbolic equations . it satisfies the burgers equation @xmath3 until its solution develops shocks . immediately after shock , the weak limit is governed by the whitham equations @xcite @xmath4 where the @xmath5 s are given by formulae ( [ lambda ] ) . equations ( [ kdvw ] ) form a @xmath6 system of hyperbolic equations . after the breaking of the solution of ( [ kdvw ] ) , the weak limit is described by a @xmath7 systems of hyperbolic equations similar to ( [ kdvw ] ) . similarly , after the solution of the @xmath7 system breaks down , the weak limit is characterized by a @xmath8 system of hyperbolic equations . in other words , for general initial data @xmath9 , one constructs the weak limit by patching together solutions of ( [ burgers ] ) , ( [ kdvw ] ) , @xmath7 , @xmath8 , etc systems in the @xmath10-@xmath11 plane . the kdv equation ( [ eq1 ] ) is just the first of an infinite sequence of equations . all these so - called higher order kdv equations can be cast in the hamiltonian form @xmath12 where @xmath13 s form a sequence of conserved functionals for the kdv equation . the small dispersive parameter @xmath14 is hidden in @xmath13 . in particular , when @xmath15 , ( [ hierarchy ] ) is the kdv equation . the solution of the higher order kdv equation ( [ hierarchy ] ) also has a weak limit as @xmath1 . as in the kdv case , this weak limit satisfies the burgers type equation @xmath16 where @xmath17 is given in ( [ cm ] ) , until the solution of ( [ mburgers ] ) forms a shock . after the breaking of the solution of ( [ mburgers ] ) , the limit is governed by equations similar to ( [ kdvw ] ) , namely , @xmath18 where @xmath19 s are given in ( [ eq18a ] ) . they will also be called the whitham equations . as in the kdv case , after the solution of ( [ mkdvw ] ) breaks down , the weak limit is described by a @xmath7 system of hyperbolic equations . in this paper , we are interested in the solution of the whitham equations for the higher order kdv ( [ hierarchy ] ) with a step - like initial function @xmath20 for such an initial function , the solution of the burgers type equation ( [ mburgers ] ) has already developed a shock at the initial time , @xmath21 . hence , immediately after @xmath21 , the whitham equations ( [ mkdvw ] ) kick in . solutions of ( [ mkdvw ] ) occupy some domains of the space - time while solutions of ( [ mburgers ] ) occupy other domains . these solutions are matched on the boundaries of the domains . equations ( [ burgers ] ) and equations ( [ mburgers ] ) are prototypes in the theory of hyperbolic conservation laws @xcite . their solutions will generally develop shocks in finite times . the solutions can be extended beyond the singularities as the entropy solutions . solutions of equations ( [ burgers ] ) or equations ( [ mburgers ] ) , in the theory of the zero dispersion limit , are not extended as weak or entropy solutions after the formation of singularities . instead , they are extended to match the whitham solutions of ( [ kdvw ] ) or ( [ mkdvw ] ) . for initial data ( [ step ] ) , the resulting solutions of the whitham equations ( [ mkdvw ] ) will be seen to be more complex than those of ( [ kdvw ] ) in the kdv case . the kdv case with the step - like initial data ( [ step ] ) was first studied by gurevich and pitaevskii @xcite . they found that it was enough to use the burgers solution of ( [ burgers ] ) and the whitham solution of ( [ kdvw ] ) to cover the whole @xmath10-@xmath11 plane , without going to the @xmath7 or @xmath8 system . namely , the space - time is divided into three parts @xmath22 the solution of ( [ burgers ] ) occupies the first and third parts , @xmath23 the whitham solution of ( [ kdvw ] ) lives in the second part , @xmath24 when @xmath25 . whether the second equation of ( [ eq9b ] ) can be inverted to give @xmath26 as a function of the self - similarity variable @xmath27 hinges on whether @xmath28 indeed , levermore @xcite has proved the genuine nonlinearity of the whitham equations ( [ kdvw ] ) , i.e. , @xmath29 for @xmath30 . for the higher order kdv ( [ hierarchy ] ) , equations ( [ mkdvw ] ) , in general , are not genuinely nonlinear , i.e. , a property like ( [ eq11 ] ) is not available . hence , solutions like ( [ eq9a ] ) and ( [ eq9b ] ) need to be modified . our construction of solutions of the whitham equation ( [ mkdvw ] ) makes use of the non - strict hyperbolicity of the equations . for kdv , it is known that the whitham equations ( [ kdvw ] ) are strictly hyperbolic , namely : @xmath31 for @xmath30 @xcite . for the higher order kdv ( [ hierarchy ] ) , different eigenspeeds of ( [ mkdvw ] ) , @xmath32 s , may coalesce in the region @xmath30 @xcite . for the higher order kdv with step - like initial function ( [ step ] ) , the space time is divided into four regions ( see figure 1 . ) @xmath33 where @xmath34 and @xmath35 are some constants . in the first and fourth regions , the solution of ( [ mburgers ] ) governs the evolution : @xmath36 the whitham solution of ( [ mkdvw ] ) lives in the second and third regions ; namely : @xmath37 when @xmath38 , and @xmath39 when @xmath40 . equations ( [ ns ] ) yield @xmath41 on a curve in the region @xmath42 . this implies the non - strict hyperbolicity of the whitham equations ( [ mkdvw ] ) for the kdv hierarchy . the @xmath43 case has been studied in @xcite . there , inequalities @xmath44 have played a crucial role in verifying that equations ( [ ns ] ) or ( [ ns2 ] ) can indeed be solved to give the solution of the whitham equations ( [ mkdvw ] ) when @xmath43 . for @xmath45 , inequalities ( [ inq ] ) are not valid any more . we therefore must use a different approach to solve the problem . the calculations are considerably more difficult than in the @xmath43 case . this is mainly because @xmath46 of ( [ q ] ) is a polynomial of degree @xmath47 when @xmath48 while it is only a quadratic polynomial when @xmath43 . the organization of the paper is as follows . in section 2 , we will study the eigenspeeds , @xmath19 s , of the whitham equations ( [ mkdvw ] ) . in section 3 , we will construct the self - similar solution of the whitham equations for the initial function ( [ step ] ) . in section 4 , we will use the self - similar solution of section 3 to construct the minimizer of a variational problem for the zero dispersion limit of the kdv hierarchy . in a subsequent publication , we will study the whitham solutions for all the other step - like initial data . in this section we define the eigenspeeds of the whitham equations for both the kdv ( [ eq1 ] ) and higher order kdv ( [ hierarchy ] ) . we first introduce the polynomials of @xmath49 for @xmath50 @xcite : @xmath51 where the coefficients , @xmath52 are uniquely determined by the two conditions @xmath53 and @xmath54 here the sign of the square root is given by @xmath55 for @xmath56 and the branch cuts are along @xmath57 and @xmath58 . in particular , @xmath59 where @xmath60 here @xmath61 and @xmath62 and @xmath63 are complete elliptic integrals of the first and second kind . @xmath62 and @xmath63 have some well - known properties @xcite . they have the expansions @xmath64 \ , \label{k}\\ e(s ) & = & \frac{\pi}{2 } [ 1 - \frac{s}{4 } - \frac{3}{64 } s^{2 } - \cdots - \frac{1}{2n-1}(\frac{1 \cdot 3 \cdots ( 2n-1 ) } { 2 \cdot 4 \cdots 2n})^{2 } s^{n } - \cdots ] \ , \label{e}\end{aligned}\ ] ] for @xmath65 . they also have the asymptotics @xmath66 \ , \label{e2 } \end{aligned}\ ] ] as @xmath67 is close to @xmath68 . furthermore , @xmath69 it immediately follows from ( [ k ] ) and ( [ e ] ) that @xmath70 the eigenspeeds of the whitham equations ( [ kdvw ] ) are defined in terms of @xmath71 and @xmath72 of ( [ p01 ] ) , @xmath73 which give @xmath74 in view of ( [ k]-[e2 ] ) , we find that @xmath75 , @xmath76 and @xmath77 have behavior : \(1 ) at @xmath78 = @xmath79 : @xmath80 \(2 ) at @xmath81 = @xmath78 : @xmath82 the eigenspeeds of the whitham equations ( [ mkdvw ] ) are @xmath83 the polynomial @xmath84 can be expressed as @xcite @xmath85 the function @xmath86 satisfies the boundary value problem for the euler - poisson - darboux equations @xmath87 \ , \label{ph3}\end{aligned}\ ] ] where @xmath88 the function @xmath89 is a quadratic polynomial in @xmath49 ; @xmath90 and @xmath91 is the solution of the boundary value problem for another version of the euler - poisson - darboux equations @xmath92 the solution of equations ( [ ph1]-[ph3 ] ) and that of ( [ q1 ] ) and ( [ q2 ] ) can be solved explicitly @xcite . in particular , the solution of ( [ q1 ] ) and ( [ q2 ] ) is @xcite @xmath93 the speeds @xmath19 s of ( [ eq18a ] ) for @xmath94 are connected to @xmath19 s for @xmath15 , which are also given by ( [ lambda ] ) . for @xmath95 , @xmath96 \ { { \partial}q(u_1 , u_2 , u_3 ) \over { \partial}u_i } + q(u_1 , u_2 , u_3 ) \ .\ ] ] we use ( [ eq18a ] ) , ( [ p ] ) and ( [ q ] ) to write @xmath97 in particular , when @xmath15 , since the corresponding @xmath98 , we obtain @xmath99 this together with ( [ mu1 ] ) proves formula ( [ q ] ) for @xmath100 . the cases for @xmath101 can be shown in the same way . @xcite 1 . @xmath102 \ , \quad \quad i , j = 1 , 2 , 3 ; i \neq j \ .\ ] ] 2 . @xmath103 + for @xmath104 and @xmath105 . the following calculations are useful in the subsequent sections . using formula ( [ q ] ) for @xmath106 and @xmath107 and formulae ( [ lambda ] ) for @xmath108 and @xmath109 , we obtain @xmath110 } m(u_1 , u_2 , u_3 ) \ , \ ] ] where @xmath111 e -(1-s)({{\partial}q \over { \partial}u_2 } + { { \partial}q \over { \partial}u_3 } ) k \ .\ ] ] we then use ( [ k3 ] ) , ( [ e3 ] ) and ( [ q1 ] ) to calculate @xmath112 \ , \label{pm } \\ { { \partial}m(u_1,u_2,u_3 ) \over { \partial}u_3 } & = & { 1 \over 2 } \ { p_2(u_1 , u_2 , u_3 ) \over u_1 - u_3 } [ e - ( 1-s ) k ] + { 3 \over 2 } \ { m(u_1 , u_2 , u_3 ) \over u_1 - u_3 } \ , \label{pm'}\end{aligned}\ ] ] where @xmath113 we next consider @xmath114 using formula ( [ q ] ) for @xmath106 and @xmath107 and formulae ( [ lambda ] ) for @xmath108 and @xmath109 , we obtain @xmath115 { { \partial}q \over { \partial}u_3 } \ , \end{aligned}\ ] ] where we have used equations ( [ q1 ] ) in the last equality . finally , we use the expansions ( [ k]-[e ] ) for @xmath116 and @xmath117 to obtain @xmath118 \ . \label{f2}\ ] ] in this section , we construct the self - similar solution of the whitham equations ( [ mkdvw ] ) when @xmath119 for the initial function ( [ step ] ) . the @xmath43 result has already been obtained in @xcite . even in the @xmath43 case , the key calculations presented here are different from those in @xcite . ( see figure 1 . ) [ main3 ] for the step - like initial data @xmath9 of ( [ step ] ) , the solution of the whitham equations ( [ mkdvw ] ) is given by @xmath120 for @xmath121 and by @xmath122 for @xmath123 , where @xmath124 , @xmath125 and @xmath126 . here , @xmath127 is uniquely determined by the equation @xmath128 and @xmath129 is uniquely given by the equation @xmath130 outside the region @xmath131 , the solution of the burgers type equation ( [ mburgers ] ) is given by @xmath132 and @xmath133 [ fig1 ] the boundaries @xmath134 and @xmath135 are called the trailing and leading edges , respectively . they separate the solutions of the whitham equations and burgers type equations . the whitham solution matches the burgers type solution in the following fashion ( see figure 1 . ) : @xmath136 at the trailing edge ; @xmath137 at the leading edge . the proof of theorem [ main3 ] is based on a series of lemmas . we first show that the solution defined by either formulae ( [ ws1 ] ) or ( [ ws2 ] ) indeed satisfies the whitham equations ( [ mkdvw ] ) @xcite . [ tsa ] 1 . the functions @xmath138 , @xmath26 and @xmath139 determined by equations ( [ ws1 ] ) give a solution of the whitham equations ( [ mkdvw ] ) as long as @xmath26 and @xmath139 can be solved from ( [ ws1 ] ) as functions of @xmath10 and @xmath11 . 2 . the functions @xmath138 , @xmath26 and @xmath139 determined by equations ( [ ws2 ] ) give a solution of the whitham equations ( [ mkdvw ] ) as long as @xmath26 can be solved from ( [ ws2 ] ) as a function of @xmath10 and @xmath11 . \(1 ) @xmath138 obviously satisfies the first equation of ( [ mkdvw ] ) . to verify the second and third equations , we observe that @xmath140 on the solution of ( [ ws1 ] ) to see this , we use ( [ a23 ] ) to calculate @xmath141 the second part of ( [ dia ] ) can be shown in the same way . we then calculate the partial derivatives of the second equation of ( [ ws1 ] ) with respect to @xmath10 and @xmath11 . @xmath142 which give the second equation of ( [ mkdvw ] ) . the third equation of ( [ mkdvw ] ) can be verified in the same way . \(2 ) the second part of lemma 3.2 can easily be proved . we now determine the trailing edge . eliminating @xmath10 and @xmath11 from the last two equations of ( [ ws1 ] ) yields @xmath143 since it degenerates at @xmath144 , we replace ( [ m23 ] ) by @xmath145 here , the function @xmath146 is also defined in ( [ f ] ) . therefore , at the trailing edge where @xmath147 , i.e. , @xmath148 , equation ( [ f1 ] ) , in view of the expansion ( [ f2 ] ) , becomes @xmath149 since @xmath150 and @xmath151 on @xmath147 because of ( [ q1 ] ) , this equation is exactly equation ( [ u * ] ) . [ u * ] equation @xmath152 has a simple zero , denoted by @xmath127 , in the region @xmath153 , counting multiplicities . furthermore , @xmath154 is positive when @xmath155 and negative when @xmath156 . we first simplify the polynomial @xmath157 of ( [ poly ] ) . in view of formula ( [ iq ] ) for @xmath46 , we use the fact that @xmath46 is symmetric in @xmath138 , @xmath26 and @xmath139 to obtain @xmath158 where @xmath159 we can then write @xmath157 as @xmath160 \ .\ ] ] denoting the function in the parenthesis of ( [ u ] ) by @xmath161 , we claim that @xmath162 for @xmath163 . obviously , @xmath164 since , @xmath165 is a positive function , this proves the second inequality of ( [ p ] ) . to prove the first inequality of ( [ p ] ) , we use formula ( [ fu ] ) to calculate @xmath166 the integral on the right can be evaluated using an iteration formula . denote this integral by @xmath167 . an integration by parts gives @xmath168 . since @xmath169 , we thus obtain @xmath170 which gives @xmath171 therefore @xmath172 we now use ( [ p ] ) to prove the existence and uniqueness of the zero of function @xmath173 , @xmath174 . first , it follows from ( [ p ] ) that @xmath173 has an odd number of zeros in @xmath153 , counting multiplicities . second , if @xmath173 has more than one zero , it must have at least three zeros . consequently , @xmath175 will have more than one zero ; so it must also have at least three zeros . repeating this argument , we see that @xmath176 must have at least three zeros . this is an impossibility since @xmath161 is a polynomial of degree @xmath177 because @xmath178 is so . therefore , @xmath173 has one and only one zero for @xmath179 when @xmath180 . in particular , the @xmath181 case proves lemma [ u * ] . [ trailing ] equation ( [ f1 ] ) has a unique solution satisfying @xmath147 . the solution is @xmath182 . the rest of equations ( [ ws1 ] ) at the trailing edge are @xmath183 and @xmath184 . having located the trailing edge , we now solve equations ( [ ws1 ] ) in the neighborhood of the trailing edge . we first consider equation ( [ f1 ] ) . we use ( [ f2 ] ) to differentiate @xmath146 at the trailing edge @xmath183 , @xmath182 @xmath185_{\xi = u^ * } > 0 \label{u+ } \ , \end{aligned}\ ] ] where in the second equality we have used ( [ q1 ] ) , ( [ divq ] ) and identities @xmath186 on @xmath147 . the inequality is a consequence of lemma [ u * ] . inequality ( [ u+ ] ) shows that equation ( [ f1 ] ) or equivalently ( [ m23 ] ) can be inverted to give @xmath26 as a decreasing function of @xmath139 @xmath187 in a neighborhood of @xmath188 . we will extend the solution ( [ b ] ) of equation ( [ m23 ] ) by decreasing @xmath139 in the region @xmath189 as far as possible . we need to evaluate the derivatives @xmath190 and @xmath191 on the solution of ( [ ws1 ] ) . it follows from ( [ m ] ) , ( [ pm ] ) and ( [ pm ] ) that @xmath192 \over { \partial}u_2 } & = & - { s k p_1(1 , u_2 , u_3 ) \over e - ( 1-s)k } \label{u22 } \ , \\ { { \partial}[\mu_2^{(m ) } - \mu_3^{(m ) } ] \over { \partial}u_3 } & = & { s k p_2(1 , u_2 , u_3 ) \over k - e } \label{u33 } \ , \end{aligned}\ ] ] on the solution of ( [ ws1 ] ) . we first study the two polynomials @xmath157 and @xmath193 of ( [ poly ] ) . [ p12 ] for each @xmath194 , the polynomial @xmath195 , as a function of @xmath26 , has only one zero in the region @xmath196 , counting multiplicities . furthermore , @xmath195 is positive when @xmath26 is on the left of this zero and negative when @xmath26 is on the right . for each @xmath197 , the polynomial @xmath198 , as a function of @xmath139 , has only one zero in the region @xmath199 , counting multiplicities . furthermore , @xmath195 is positive when @xmath139 is on the left of this zero and negative when @xmath139 is on the right . we will prove the first part of the lemma ; the second part follows from @xmath200 . the proof of the first part is similar to the proof of lemma [ u * ] . we will go through it briefly . we first have @xmath201 for @xmath202 . the second inequality immediately follows from formula ( [ poly ] ) for @xmath157 . the first inequality is derived from a formula similar to ( [ uk ] ) . the rest of the proof is the same as the proof of lemma [ u * ] . we now continue to extend the solution ( [ b ] ) of equation ( [ m23 ] ) in the region @xmath189 as far as possible . when @xmath26 and @xmath139 are close to @xmath127 , because of ( [ u+ ] ) , we have @xmath203 and @xmath204 on the solution of ( [ m23 ] ) . these along with ( [ u22 ] ) and ( [ u33 ] ) show that @xmath205 when @xmath26 and @xmath139 are close to @xmath127 . [ p12 > ] inequalities ( [ inq ] ) hold on the solution of equation ( [ m23 ] ) as long as @xmath189 . we first prove the first inequality of ( [ inq ] ) . in view of ( [ m ] ) , the solution of equation ( [ m23 ] ) is also governed by an equivalent equation @xmath206 . for each @xmath207 , we study the zero of the @xmath26-variable function @xmath208 in the interval @xmath209 . we obtain from formula ( [ m1 ] ) for @xmath210 that @xmath211 and @xmath212 . @xmath208 is a decreasing function of @xmath26 when @xmath26 is on the immediate right of @xmath139 . to see this , we note that @xmath213 for @xmath214 according to lemma [ u * ] . this and ( [ pm ] ) prove that @xmath210 is decreasing for @xmath26 on the immediate right of @xmath139 because @xmath215 for @xmath216 ; so @xmath217 for such @xmath26 . therefore , @xmath208 has a @xmath26-zero in the interval @xmath209 when @xmath207 . because of the uniqueness of the @xmath26-zero of @xmath218 according to lemma [ p12 ] , we conclude that @xmath208 has only one zero and that this zero is on the right of the zero of @xmath157 . hence , the zero of @xmath210 is exactly given by @xmath219 and @xmath157 is negative on the solution . this proves the first inequality of ( [ inq ] ) . we now prove the second inequality of ( [ inq ] ) by contradiction . suppose it first fails at @xmath220 and @xmath221 , where @xmath222 ; i.e. , @xmath223 it then follows from ( [ u22 ] ) and ( [ u33 ] ) that solution ( [ b ] ) of equation ( [ m23 ] ) has a zero derivative at @xmath224 ; i.e. , @xmath225 hence , @xmath226 where the first term vanishes because of @xmath227 and the second term is negative according to lemma [ p12 ] . in view of @xmath228 , this implies that @xmath229 when @xmath139 is on the immediately right of @xmath221 . that contradicts ( [ 2 ] ) . this proves the second inequality of ( [ inq ] ) . it follows from ( [ u22 ] ) , ( [ u33 ] ) and lemma 3.6 that @xmath230 \over { \partial}u_2 } > 0 \ , \quad { { \partial}[\mu_2^{(m ) } - \mu_3^{(m ) } ] \over { \partial}u_3 } > 0\ ] ] on the solution of ( [ m23 ] ) . solution ( [ b ] ) of equation ( [ m23 ] ) can then be extended as a decreasing function of @xmath139 as long as @xmath189 . there are two possibilities : ( 1 ) @xmath26 touches @xmath68 before or simultaneously as @xmath139 reaches @xmath231 and ( 2 ) @xmath139 touches @xmath231 before @xmath26 reaches @xmath68 . it follows from ( [ le ] ) and ( [ q ] ) that @xmath232 this shows that ( 1 ) is impossible . hence , @xmath139 will touch @xmath231 before @xmath26 reaches @xmath68 . when this happens , equation ( [ m23 ] ) becomes equation ( [ mid ] ) . [ u2 * ] equation ( [ mid ] ) has a simple zero in the region @xmath196 , counting multiplicities . denoting the zero by @xmath129 , then @xmath233 is positive for @xmath234 and negative for @xmath235 . furthermore , @xmath236 where @xmath237 is the unique zero of @xmath238 . we use ( [ m ] ) and ( [ pm ] ) to prove the lemma . in equation ( [ m ] ) , @xmath239 and @xmath240 are all positive for @xmath241 in view of ( [ ke ] ) . denoting the unique zero of @xmath238 by @xmath237 , it then follows from lemma [ p12 ] that @xmath242 when @xmath243 and @xmath244 when @xmath245 . since @xmath246 of ( [ m1 ] ) vanishes at @xmath247 and is positive at @xmath248 in view of ( [ k]-[e2 ] ) , we conclude from the derivative ( [ pm ] ) that @xmath246 has a simple zero in @xmath249 . this zero is exactly @xmath129 and the rest of the theorem can be proved easily . having solved equation ( [ m23 ] ) for @xmath26 as a decreasing function of @xmath139 for @xmath207 , we turn to equations ( [ ws1 ] ) . because of ( [ dia ] ) and ( [ dia2 ] ) , the third equation of ( [ ws1 ] ) gives @xmath139 as a decreasing function of @xmath27 for @xmath250 , where @xmath124 and @xmath125 . consequently , @xmath26 is an increasing function of @xmath27 in the same interval . [ sws1 ] the last two equations of ( [ ws1 ] ) can be inverted to give @xmath26 and @xmath139 as increasing and decreasing functions , respectively , of the self - similarity variable @xmath27 in the interval @xmath250 . we now turn to equations ( [ ws2 ] ) . we first use ( [ k3 ] ) , ( [ e3 ] ) and ( [ lambda ] ) to calculate the derivative of @xmath251 of ( [ q ] ) @xmath252 { { \partial}^2 q \over { \partial}u_2 ^ 2 } + { 1 \over 2 } { { \partial}\lambda_2 \over { \partial}u_2 } { { \partial}q \over { \partial}u_2 } { \nonumber}\\ & = & { 2(u_2 - 1 ) s k \over e - ( 1-s ) k } \ { { \partial}^2 q \over { \partial}u_2 ^ 2 } + [ { 2 s k \over e - ( 1-s ) k } + 1 - { e^2 - ( 1-s)k^2 \over ( e - ( 1-s)k)^2 } ] { { \partial}q \over { \partial}u_2 } { \nonumber}\\ & > & - { 2s k \over e - ( 1-s)k } \ [ ( 1 - u_2 ) { { \partial}^2 q \over { \partial}u_2 ^ 2 } - { { \partial}q \over { \partial}u_2 } ] \ , \label { * } \end{aligned}\ ] ] where in the inequality we have used @xmath253 , which is a consequence of ( [ ke ] ) . the polynomial in the parenthesis of ( [ * ] ) is connected to @xmath254 ; indeed , @xmath255 \ .\ ] ] this follows from the identity @xmath256 to see this , taking the derivative of ( [ identity3.27 ] ) and using formula ( [ poly ] ) for @xmath157 and equations ( [ q1 ] ) for @xmath46 yield ( [ * * ] ) . to prove ( [ identity3.27 ] ) , we use the integral formula ( [ iq ] ) for @xmath46 to calculate both sides of the identity . the left equals @xmath257 \left [ \int_{-1}^1 \left ( ( 1+\nu ) + ( 1-\nu ) u_2 \right)^{m-1 } ( 1-\nu^2)^{-1/2 } d\nu \right ] \,.\ ] ] the right is @xmath258 \left [ \int_{-1}^1 \left ( ( 1+\nu ) + ( 1-\nu ) u_2 \right)^{m-1 } ( 1-\nu^2)^{-1/2 } d\nu \right ] \ .\ ] ] both sides are equal in view of an easy identity @xmath259 we have therefore proved identity ( [ identity3.27 ] ) . by lemma [ p12 ] , @xmath254 is negative for @xmath260 , where @xmath237 is the unique zero of @xmath157 . since @xmath261 according to lemma [ u2 * ] , we conclude from ( [ * ] ) and ( [ * * ] ) that @xmath262 on the solution of ( [ ws2 ] ) when @xmath234 . hence , the second equation of ( [ ws2 ] ) can be solved for @xmath26 as an increasing function of @xmath27 as long as @xmath263 . when @xmath26 reaches @xmath68 , we have @xmath264 we have therefore proved the following result . [ sws2 ] the second equation of ( [ ws2 ] ) can be inverted to give @xmath26 as an increasing function of @xmath27 in the interval @xmath265 . we are ready to conclude the proof of theorem 3.1 . the burgers type solutions ( [ bs1 ] ) and ( [ bs2 ] ) are trivial . according to lemma [ sws1 ] , the last two equations of ( [ ws1 ] ) determine @xmath26 and @xmath139 as functions of @xmath27 in the region @xmath250 . by the first part of lemma [ tsa ] , the resulting @xmath138 , @xmath26 and @xmath139 satisfy the whitham equations ( [ mkdvw ] ) . furthermore , the boundary conditions ( [ tr1 ] ) and ( [ tr2 ] ) are satisfied at the trailing edge @xmath266 . similarly , by lemma [ sws2 ] , the second equation of ( [ ws2 ] ) determines @xmath26 as a function of @xmath27 in the region @xmath267 . it then follows from the second part of lemma [ tsa ] that @xmath138 , @xmath26 and @xmath139 of ( [ ws2 ] ) satisfy the whitham equations ( [ mkdvw ] ) . they also satisfy the boundary conditions ( [ le1 ] ) and ( [ le2 ] ) at the leading edge @xmath268 . we have therefore completed the proof of theorem [ main3 ] . the zero dispersion limit of the solution of the higher order kdv equation ( [ hierarchy ] ) with step - like initial function ( [ step ] ) is also determined by a minimization problem with constraints @xcite @xmath269 \psi(\eta ) d \eta \ } \ .\ ] ] in this section , we will use the self - similar solution of section 3 to construct the minimizer for @xmath119 . the @xmath43 result has already been obtained in @xcite . even in the @xmath43 case , the key calculations presented here are different from those in @xcite . 1 . for @xmath273 , @xmath274 2 . for @xmath275 , @xmath276 where @xmath71 and @xmath277 are defined in ( [ eq15 ] ) and @xmath26 and @xmath139 are determined by equations ( [ ws1 ] ) . 3 . for @xmath278 , @xmath279 where @xmath26 is determined by ( [ ws2 ] ) . 4 . for @xmath280 , @xmath281 we extend the function @xmath282 defined on @xmath283 $ ] to the entire real line by setting @xmath284 for @xmath285 and taking @xmath282 to be odd . in this way , the operator @xmath286 is connected to the hilbert transform @xmath287 on the real line @xcite : @xmath288 we verify case ( 4 ) first . clearly @xmath284 satisfies the constraint ( [ con ] ) . we now check the variational conditions ( [ con1]-[con2 ] ) . since @xmath289 , @xmath290 where the inequality follows from @xmath280 and @xmath291 . hence , variational conditions ( [ con1]-[con2 ] ) are satisfied . next we consider case ( 1 ) . we write @xmath292 as the real part of @xmath293 for real @xmath294 , where @xmath295 \over \sqrt{\eta^2 -1 } } \ .\ ] ] the function @xmath296 is analytic in the upper half complex plane @xmath297 and @xmath298 for large @xmath299 in view of the expansion ( [ eq15 ] ) for @xmath277 . hence , @xmath300 = x - 4^m ( 2m+1 ) t \eta^{2m}$ ] on @xmath291 , where @xmath287 is the hilbert transform @xcite . we then have for @xmath291 @xmath301 which shows that the variational conditions ( [ con1 ] ) and ( [ con2 ] ) are satisfied . to prove ( [ con ] ) , we first claim that @xmath302 for @xmath291 . to see this , we use ( [ tr ] ) and ( [ q ] ) to calculate @xmath303 and ( [ p ] ) and ( [ q ] ) to evaluate @xmath304 . the left hand side of ( [ 1 ] ) equals @xmath305 { \nonumber}\\ & = 2(\eta^2 -1 ) [ c_m u(\eta^2 , 1 ) - c_m u(u^ * , 1 ) ] + 2(\eta^2 - u^ * ) c_m u(u^ * , 1 ) { \nonumber}\\ & = 2[(1-u^ * ) c_mu(u^ * , 1 ) - ( 1-\eta^2 ) c_m u(\eta^2,1 ) ] \ , \label{2 } \end{aligned}\ ] ] where in the first equality we have used ( [ divq ] ) and the identity ( cf . ( 3.28 ) of @xcite ) @xmath306 in view of ( [ u ] ) , we have @xmath307 = c_m[(1-\eta^2 ) { { \partial}u(\eta^2 , 1 ) \over { \partial}\eta^2 } - u(\eta^2 , 1 ) ] = p_1(1 , \eta^2 , \eta^2 ) \ .\ ] ] this derivative is positive when @xmath308 and negative when @xmath309 because @xmath157 is so according to lemma [ u * ] . therefore , @xmath310 has a maximum at @xmath311 ; i.e. , @xmath312 for @xmath313 . this together with ( [ 2 ] ) proves ( [ 1 ] ) . the function @xmath317 is analytic in @xmath297 and @xmath318 for large @xmath299 in view of the asymptotics ( [ eq16 ] ) for @xmath71 and @xmath277 . hence , taking the imaginary part of @xmath317 yields @xmath319 \eta \over \sqrt{(1 - \eta^2 ) ( u_2 - \eta^2)(\eta^2 - u_3 ) } } & \sqrt{u_3 } < \eta < \sqrt{u_2 } \\ x - 4^m ( 2m+1 ) t \eta^{2 m } & \sqrt{u_2 } < \eta < 1 \ . \end{matrix } \right.\ ] ] we then have @xmath320 \mu \over \sqrt{(1 - \mu^2 ) ( u_2 - \mu^2 ) ( \mu^2 - u_3 ) } } d \mu & \sqrt{u_3 } < \eta < \sqrt{u_2 } \\ x \eta - 4^m t \eta^{2m+1 } & \sqrt{u_2 } < \eta < 1 \ , \end{matrix } \right.\ ] ] where we have used @xmath321 \mu \over \sqrt{(1 - \mu^2 ) ( u_2 - \mu^2 ) ( \mu^2 - u_3 ) } } d \mu = 0 \ , \ ] ] which is a consequence of ( [ eq17 ] ) for @xmath71 and @xmath277 . to verify ( [ con ] ) , we derive an integral formula for @xmath282 . we use ( [ pm0 ] ) and ( [ ws1 ] ) to calculate @xmath322 integrating yields @xmath323 where @xmath324 is defined in figure 1 . the polynomial @xmath71 of ( [ eq15 ] ) is linear in @xmath325 and has a zero for @xmath326 because of ( [ eq17 ] ) . @xmath71 must be positive for @xmath327 and negative for @xmath328 . this combined with ( [ int ] ) proves @xmath314 ; so ( [ con ] ) is verified . we now continue to verify the variational conditions ( [ con1 ] ) and ( [ con2 ] ) . again , we use ( [ pm0 ] ) and ( [ ws1 ] ) to calculate @xmath329\mu \over \sqrt{(1 - \mu^2)(u_2(x , t ) - \mu^2)(\mu^2 - u_3 ^ 2(x , t ) ) } } \right ) \\ = { - \mu p_0(\mu^2 , 1 , u_2(x , t ) , u_3(x , t ) ) \over \sqrt{(1 - \mu^2)(u_2(x , t ) - \mu^2)(\mu^2 - u_3 ^ 2(y , t ) ) } } \ . \end{gathered}\ ] ] integrating yields @xmath330\mu \over \sqrt{(1 - \mu^2)(u_2(x , t ) - \mu^2)(\mu^2 - u_3 ^ 2(x , t ) ) } } \\ = - \int_{\chi(\mu^2 ) t } ^x { \mu p_0(\mu^2 , 1 , u_2(y , t ) , u_3(y , t ) ) \over \sqrt{(1 - \mu^2)(u_2(y , t ) - \mu^2)(\mu^2 - u_3 ^ 2(y , t ) ) } } d y \end{gathered}\ ] ] for @xmath331 . the single integral in ( [ lpsi ] ) can then be written as a double integral . after interchanging integrals and using ( [ eq17 ] ) for @xmath71 , the double integral is simplified as @xmath332 the polynomial @xmath71 is linear in @xmath333 and has a zero for @xmath334 . in view of ( [ eq17 ] ) for @xmath71 , we must have @xmath335 for @xmath336 . hence , the integral in ( [ lpsi ] ) is positive and this verifies the variational conditions ( [ con1 ] ) and ( [ con2 ] ) . we write @xmath338 for real @xmath294 , where @xmath339 \over \sqrt{(\eta^2 -1)(\eta^2 - u_2 ) } } \ .\ ] ] the function @xmath340 is analytic in @xmath297 and @xmath341 for large @xmath299 in view of the asymptotics ( [ eq16 ] ) for @xmath71 and @xmath277 . hence , taking the imaginary part of @xmath340 yields @xmath342 we then have @xmath343 where we have used @xmath344 which is a consequence of ( [ eq17 ] ) for @xmath71 and @xmath277 . * acknowledgments . * we thank tamara grava for some of the ideas leading to the proof of lemma [ u * ] . v.p . was supported in part by nsf grant dms-0135308 . t. was supported in part by nsf grant dms-0404931 . lax , c.d . levermore and s. venakides , `` the generation and propagation of oscillations in dispersive ivps and their limiting behavior '' in important developments in soliton theory 1980 - 1990 , t. fokas and v.e . zakharov eds . , springer - verlag , berlin ( 1992 ) . tian , `` on the initial value problem of the whitham averaged system '' , in singular limits of dispersive waves , n. ercolani , i. gabitov , d. levermore and d. serre eds . , nato arw series , series b : physics vol . 320 , plenum , new york ( 1994 ) , 135 - 141 .	we study the whitham equations for all the higher order kdv equations . the whitham equations are neither strictly hyperbolic nor genuinely nonlinear . we are interested in the solution of the whitham equations when the initial values are given by a step function . [ multiblock footnote omitted ]
the study of the properties of unstable nuclei both theoretically and experimentally is nowadays one of the most active and fruitful research lines in nuclear physics . nuclear systems with very unusual @xmath1 ratios are proper candidates to get insight into the nuclear interaction and the impact that the associated dynamics might have in low - energy observables like the ground - state deformation and derived quantities like moments of inertia or vibrational excitation energies @xcite . the understanding of the properties of those nuclei also has important consequences in the understanding of other physical processes like stellar nucleosynthesis of heavy elements . in particular , the neutron - rich sr , zr , and mo isotopes with mass numbers @xmath2 are of special interest for various reasons . one of them is the role that these isotopes play in the nucleosynthesis of heavy nuclides in the astrophysical r process . their masses and decay properties are an essential input to model the path , the isotopic abundances and the time scale of the r process in a reliable way @xcite . in addition , from the nuclear structure point of view , this region is characterized by a strong competition between various shapes , giving rise to shape instabilities that lead to coexisting nuclear shapes , as well as to sudden shape transitions in isotopic chains @xcite . it has been shown that the ground states of sr , zr and mo isotopes with @xmath3 ranging from the magic number @xmath4 up to @xmath5 are weakly deformed , but they undergo a shape transition from nearly spherical to well deformed prolate ( or oblate ) deformations as @xmath6 is approached and crossed . evidence for such an abrupt shape change includes @xmath7 lifetime measurements @xcite as well as quadrupole moments for rotational bands @xcite . signatures of triaxiality in mo isotopes from spectroscopic studies of high - spin states have also been identified @xcite . heavier sr and zr ( @xmath8 ) isotopes display an axially symmetric well deformed shape . above this region , it has been suggested that the @xmath9 shell closure might be quenched far from stability . this quenching has been predicted by different models @xcite , but still weak experimental evidence has been found . masses ( nuclear binding energies ) and charge radii are considered among the most sensitive observables to explore the structural evolution in nuclei far from the valley of stability , where other types of spectroscopic experiments are prohibitive because of the low production rates and short lifetimes of these unstable isotopes . discontinuities in the systematics of particular mass regions suggest a change in the structure of the ground states among the configurations competing for the lowest energy . drastic changes in the mean square ( ms ) radius of the charge distributions in regions of transitional nuclei may also be indicators of structural changes related to the nuclear deformation . in this work we study this phenomenology in the framework of the self - consistent hartree - fock - bogoliubov ( hfb ) approximation based on the finite range and density dependent gogny interaction @xcite with the parametrization d1s @xcite . the hfb wave functions are expanded in a harmonic oscillator basis containing enough number of shells to achieve convergence of the results for all the nuclei studied . the quadrupole deformations at equilibrium are obtained self - consistently by a minimization procedure . nevertheless , constrained calculations have also been performed to generate potential energy curves ( pec ) or potential energy surfaces ( pes ) to study the evolution with the number of nucleons of the various shapes associated to the different minima displayed by the pecs or pess . this has been accomplished preserving axial symmetry first , and allowing afterward for triaxiality in the most relevant isotopes where shape transitions take place . hence , we have constrained the mean value of the quadrupole operators @xmath10 and @xmath11 ( in the axial case only the former is constrained ) , @xmath12 @xmath13 in the triaxial figures we plot @xmath14 planes , where @xmath15 and @xmath16 is the angle defined as @xmath17 @xcite . with this definition an axially symmetric prolate mass distribution has @xmath18 , whereas the corresponding oblate one has @xmath19 . the quadrupole deformation parameter @xmath20 is defined in terms of the mass quadrupole moment @xmath21 and ms radius @xmath22 , @xmath23 for the description of odd-@xmath0 nuclei , we have extended the hfb formalism using blocking techniques . the blocked hfb wave function of the odd - a system is in general given by @xmath24 where @xmath25 is a reference even - even hfb vacuum ( @xmath26 ) and @xmath27 is a quasiparticle creation operator . the index @xmath28 stands for the quasiparticle quantum numbers characterizing the blocked state ( angular momentum projection @xmath29 and parity in the case of axial symmetry ) . the ground state of the odd nucleus is determined by finding the blocked state that minimizes the total energy . in the present study we use the equal filling approximation ( efa ) , a prescription widely used in mean - field calculations to treat the dynamics of odd nuclei in a time - reversal invariant way . in this approximation the unpaired nucleon is treated in an equal footing with its time - reversed state by sitting half a nucleon in a given orbital and the other half in the time - reversed partner . this procedure has been recently justified microscopically , showing that the efa can be described in terms of a mixed state density operator and the equations to be solved are a direct consequence of the variational principle over the energy of such mixed state @xcite . it has also been shown that the efa and the exact blocking procedure are both strictly equivalent when the time - odd fields of the energy density functional are neglected @xcite . thus , efa is sufficiently precise for most practical applications . in this work , we present results for one - neutron ( @xmath30 ) and two - neutron ( @xmath31 ) separation energies , which can be easily calculated from the binding energies @xmath32 , @xmath33 charge radii and their differences are crucial quantities to study the systematics of the nuclear - shape changes as they can be measured with remarkable precision using laser spectroscopic techniques . they are obtained theoretically by folding the proton distribution of the nucleus with the the finite size of the protons and the neutrons . the ms radius of the charge distribution in a nucleus can be expressed as @xmath34 where @xmath35 is the ms radius of the point proton distribution in the nucleus @xmath36 @xmath37 @xmath38 and @xmath39 @xmath38 are the ms radii of the charge distributions in a proton and a neutron , respectively . @xmath40 is a small correction due to the center of mass motion , which is evaluated according to ref . it is worth noticing that the most important correction to the point proton ms nuclear radius , coming from the proton charge distribution @xmath41 , vanishes when isotopic differences are considered , since it does not involve any dependence on @xmath3 . the variations of the charge radii in isotopic chains are related to deformation effects . other structure effects like pairing or spin - orbit couplings have been considered @xcite , but in this work they are fixed by the gogny interaction . for an axially symmetric static quadrupole deformation @xmath20 the increase of the charge radius with respect to the spherical value is given to first order by @xmath42 where usually @xmath43 is taken from the droplet model . then , the measured radii have been used to estimate the changes in quadrupole deformation @xcite . on the contrary , in this work we analyze the effect of the quadrupole deformation on the charge radii from a microscopic self - consistent approach . ( a ) , @xmath31 ( b ) , and @xmath30 ( c ) in sr isotopes compared to experimental data from ref . @xcite for masses and from ref . @xcite for radii . results for prolate , oblate , and spherical minima are displayed with different symbols ( see legend ) . open circles correspond to ground - state results . , width=302 ] , but for zr isotopes . experimental data for masses and radii are from refs . @xcite and @xcite , respectively.,width=302 ] , but for mo isotopes . experimental data for masses and radii are from refs . @xcite and @xcite , respectively.,width=302 ] the pecs of the neutron - rich isotopes of sr , zr , and mo from @xmath4 to @xmath44 have been calculated with the gogny - d1s interaction . our results fully agree with those of ref . @xcite that were obtained in the same framework . the main features of the shape evolution in the axial case can be summarized as follows : the isotopes with @xmath45 show a sharp pec around the spherical minimum that becomes rather shallow at @xmath46 . isotopes with @xmath47 are already deformed with oblate and prolate minima very close in energy . in the case of sr isotopes the ground state is prolate , for zr isotopes both oblate and prolate minima are found at about the same energy , while for mo isotopes the ground state is oblate . beyond @xmath47 the shapes become stable and for heavier isotopes we obtain basically similar results to @xmath47 . it is also worth mentioning the incipient emergence of a spherical solution in the heavier isotopes . the shape change at @xmath5 has been predicted to a different extent by various theoretical models . global calculations within the finite - range droplet model @xcite with single - particle states obtained from folded yukawa predict prolate quadrupole deformations increasing smoothly from @xmath4 up to @xmath48 , and then jumping suddenly to large deformations between @xmath49 and @xmath47 . at @xmath50 the shapes stabilize until a transition to oblate shapes is predicted at @xmath51 in sr and zr , and at @xmath52 in mo . triaxial calculations within this approach were carried out in ref . @xcite , showing that only mo isotopes have a tendency to triaxiality in this mass region . the approach followed in ref . @xcite , where pess were studied within a finite - range liquid drop model modified by shell corrections taken from deformed woods - saxon potentials , suggests an oblate - prolate shape coexistence in sr and zr isotopes from @xmath53 with prolate ground states . mo isotopes display in this case a soft behavior that develops triaxiality at @xmath54 . other calculations including rotational states in terms of the total routhian surface , using non - axial wood - saxon potentials @xcite , predict two coexisting prolate and oblate minima for @xmath55zr isotopes , where the prolate ground state becomes oblate beyond @xmath56zr . the same calculations predict a @xmath16-soft triaxial minimum for @xmath57mo . planes for @xmath58sr isotopes with the gogny - d1s interaction . the contour lines extend from the minimum up to 2 mev higher in steps of 0.25 mev.,width=302 ] , but for @xmath59zr isotopes.,width=302 ] , but for @xmath60mo isotopes.,width=302 ] zr ( @xmath61 ) as a function of the axial quadrupole moment @xmath21 ( b ) . the fermi level is plotted by a thick dotted line.,width=283 ] relativistic mean - field calculations @xcite also show increasing prolate deformations from @xmath4 up to @xmath47 . in the region between @xmath47 and @xmath62 the deformations are stable for both sr and zr , whereas they change to oblate at @xmath63 in mo . triaxial pess calculated from skyrme hf+bcs were already published in ref . @xcite , where the shape evolution along kr , sr , zr , and mo isotopes was studied . finally , large scale studies of quadrupole correlation energies and ms charge radii , based on both skyrme - sly4 @xcite and gogny - d1s @xcite effective interactions have been carried out recently . in summary , most of the theoretical nuclear models predict increasing deformations up to @xmath47 , where oblate and prolate shapes are developed and exhibit minima at close energies . which of them becomes the ground state depends on subtle details of the calculation . at variance with the calculations mentioned above , in the present work we also consider odd-@xmath0 nuclei in the framework of the efa . in the next figures we can see the results obtained for @xmath64 in ( a ) , for @xmath31 energies in ( b ) , and for @xmath30 energies in ( c ) . [ fig_sr_r2 ] corresponds to sr , fig . [ fig_zr_r2 ] to zr , and fig . [ fig_mo_r2 ] to mo isotopes . in the three figures , results are shown as a function of the neutron number for all possible values of @xmath3 , even or odd , in the range of interest . experimental data have been taken from the mass measurements @xcite and from laser spectroscopy experiments carried out at isolde / cern and at the igisol facility of the university of jyvskyl @xcite . focusing on the experimental data , a consistent picture emerges . besides the abrupt decrease of @xmath31 and @xmath30 at @xmath4 corresponding to the shell closure , the evolution of the @xmath31 and @xmath30 along the isotopic chains shows a change in the tendency around @xmath47 in sr and zr isotopes . this suggests a change in the ground - state shape of these isotopes . on the other hand , the chain of mo isotopes shows a smoother behavior . these observations are confirmed by laser spectroscopy experiments measuring the nuclear ms radius . in this case , the shape change in sr and zr is observed in the form of a sudden increase of the ms charge radii at @xmath65 . again , the mo isotopic chain @xcite displays a smooth behavior as the neutron number increases . in these figures we also see the calculated values using the oblate ( down triangles ) , prolate ( up triangles ) and spherical ( dots ) shapes corresponding to the local minima in the pecs . we also plot by open circles the theoretical values that correspond to the ground states of the corresponding isotopes . in general , the measured @xmath31 are reproduced reasonably well . the calculated shell gap at @xmath4 is larger than observed , but this is a well known feature related to any mean - field calculation . the discrepancy can be reduced considering dynamical correlations beyond mean field @xcite . between @xmath66 and @xmath49 , the @xmath31 energies are underestimated by the calculations , while they are much better reproduced beyond @xmath47 . in our calculations a change in the tendency is observed at @xmath5 , which is more pronounced in sr and zr isotopes and almost disappears in mo isotopes . however , the shoulder is less apparent than the experimental one due to the underestimation of the data below @xmath47 . notice that the open circles at @xmath67 in @xmath31 and @xmath30 do not coincide with any specific deformation ( oblate or prolate ) because they correspond to the difference between the prolate @xmath67 and the oblate @xmath49 in @xmath30 and @xmath68 in @xmath31 . in the case of @xmath30 energies , the amplitude of the odd - even staggering in @xmath30 is well reproduced by the calculations indicating the validity of our theoretical description of odd - a nuclei . in general , the agreement is fairly good below @xmath4 . then , the calculations underestimate the measured @xmath30 values between @xmath4 and @xmath47 , being the net effect a displacement to slightly lower energies . the agreement improves substantially for heavier isotopes . the evolution of the nuclear charge radii can be seen in panels ( a ) of figs . [ fig_sr_r2 ] , [ fig_zr_r2 ] and [ fig_mo_r2 ] , where the radius , relative to that of the @xmath4 isotope , is plotted as a function of neutron number . results for both even - even and odd - a isotopes are included in the plot . it is worth pointing out the almost negligible odd - even staggering observed both in the theoretical predictions and the experimental data . in fig . [ fig_sr_r2 ] , for sr isotopes , we can see that the calculations follow nicely the measurements . there is a smooth increase of @xmath69 up to @xmath49 , then a sudden jump occurs at @xmath47 and again the increase is smooth for heavier isotopes . the encircled symbols indicate the shapes corresponding to the ground states obtained in our calculations . we can see that the lighter isotopes are spherical , then they change into oblate very smoothly and at about @xmath47 they become prolate . the observed jump corresponds to the transition from the oblate to the prolate shape since the oblate deformation is placed at @xmath70 while the prolate one appears with a different magnitude ( @xmath71 ) . [ fig_zr_r2](a ) shows similar results for zr isotopes . here , the spherical shapes account for the behavior of @xmath69 up to @xmath48 . from @xmath48 up to @xmath47 there is a smooth transition to oblate shapes that become the ground states and reproduce quite well the experiment . above @xmath47 we obtain prolate ground states with radii in agreement with the observed jump . for heavier isotopes ( beyond @xmath72 ) we obtain again oblate shapes but there is no information in this region . finally , fig . [ fig_mo_r2](a ) for mo isotopes shows that the lightest isotopes are spherical changing into oblate shapes and increasing the @xmath69 very smoothly . in general , we observe that the calculations from @xmath73 with spherical ground states underestimate the data for the nuclear radius in the three isotopic chains . one should notice that these spherical solutions are not very sharp but shallow minima as @xmath3 increases . thus , a possible explanation for the discrepancy in @xmath69 could be that configuration mixing plays an important role in these isotopes and the actual ground state will have contributions not only from the spherical configuration , but also from neighbor deformed states that will increase slightly the charge radii . in the case of sr isotopes , the sudden change in @xmath69 occurs experimentally between @xmath67 and @xmath47 , whereas theoretically it appears between @xmath49 and @xmath67 . similarly in zr isotopes we obtain the change between @xmath74 , while experimentally is observed between @xmath75 . we do not think this discrepancy is significant since it is related to the subtle competition between prolate and oblate shapes . we should notice that in the isotopes where the shape is changing we get practically degenerate energies for oblate and prolate deformations and thus , tiny changes in the details of the calculations can lead to a different ground state . in particular , if the ground state in @xmath67 in sr were oblate instead of prolate ( they are separated by less than 0.5 mev in the calculations ) we will get agreement with the experiment . similarly , in zr isotopes , if @xmath47 were prolate instead of oblate ( separated by 0.2 mev in the calculations ) we would match the experimental jump . in fig . [ fig_mo_r2](a ) we also recognize a difficulty in the reproduction of the data in the heavier isotopes , since the oblate shapes , which are the ground states , underestimate them . however , we can see that the prolate shapes , which are close in energy , agree with the data . to get a further insight into the reason for this discrepancy and for the fact that the observed jump at @xmath47 in sr and zr almost vanishes in mo , we have performed triaxial calculations for the critical isotopes around @xmath47 . similar calculations can also be found in ref . we can see in figs . [ fig_sr_tri ] , [ fig_zr_tri ] and [ fig_mo_tri ] the @xmath14 plots for @xmath7658 , 60 and 62 in sr ( fig . [ fig_sr_tri ] ) and zr ( fig . [ fig_zr_tri ] ) , as well as for @xmath7658 , 60 , 62 , 64 , 66 and 68 in mo isotopes ( fig . [ fig_mo_tri ] ) . in these figures we can see that in sr and zr isotopes the transition from oblate to prolate at @xmath47 is manifest , suddenly changing from deformations with @xmath77 b in the oblate sector to @xmath78 b in the prolate one . on the contrary , in fig . [ fig_mo_tri ] for mo isotopes we see that the oblate shape at @xmath49 becomes gradually triaxial as @xmath3 increases . an island of triaxiality is apparent from @xmath47 up to @xmath44 . we have calculated the charge radii corresponding to these triaxial configurations that become ground states and have added them in fig . [ fig_mo_r2](a ) with open squares . the new @xmath69 values for these isotopes are now very close to the axial - prolate values and agree very nicely with the experiment . one should notice that one important ingredient for the agreement achieved is that the location of the quadrupole @xmath79-value for the triaxial minima is much closer to the axial prolate minima than to the oblate ones , which are lower . to further illustrate the emergence and competition of deformed configurations in this mass region , we show in fig . [ fig_zr_sp ] , for the case of @xmath80zr ( @xmath8140 , @xmath7660 ) , the proton and neutron single - particle energies ( spe ) as functions of the axial quadrupole moment @xmath21 . fermi levels are plotted with thick dotted lines . these diagrams help us to identify the regions of low level density , which favor the onset of deformation ( jahn - teller effect ) , as well as to stress the important role of the interplay between the proton @xmath82 and the neutron @xmath83 orbitals ( federman - pittel effect ) to generate deformed configurations @xcite . in the plot for the proton spe we observe an energy gap below the fermi level at @xmath84 b , which favors the onset of prolate deformation in sr isotopes . on the other hand , above the fermi energy , the low level density on the oblate sector favors oblate configurations in mo isotopes . in the case of neutrons the energy gap below the fermi level at @xmath85 b favors oblate shapes in lighter isotopes ( @xmath86 ) , whereas the energy gap above the fermi level at @xmath84 b favors prolate shapes in heavier isotopes ( @xmath87 ) . thus , these simple ideas offer a qualitative understanding of the various mechanisms leading to deformation in this mass region . we have used self - consistent hfb calculations based on the interaction gogny - d1s to study neutron separation energies and charge radii in neutron - rich sr , zr , and mo isotopic chains . our primary aim has been to search for signatures of structural changes , and more specifically , for shape transitions in these observables . we have found these correlations and specifically a remarkable connection between shape transitions and sudden changes in the behavior of the isotopic dependence of the nuclear charge radii . the different sensitivities of @xmath31 and @xmath69 can be understood because shape transitions take place in this mass region through isotopes characterized by shape coexistence , where the energies of the various shapes are almost degenerate . thus , neutron separation energies are not particularly sensitive to these changes . on the other hand , the sensitivity to shape transitions is enhanced for charge radii , specially when the transition takes place suddenly between nuclear shapes at different absolute values of the deformation parameter . as compared to sr and zr isotopes , we have found that mo isotopes exhibit a smoother increase in the charge radii with the number of neutrons , which is in good agreement with the experimental data . triaxial degrees of freedom are required to get this agreement beyond @xmath47 . from triaxial calculations we have shown that sr and zr isotopes suffer a sharp transition from oblate to prolate shapes at @xmath47 . on the other hand mo isotopes display a smooth transition through a wide region of triaxiality . in summary , we have shown the ability of the isotopic differences in nuclear charge radii to signal structural changes related to deformation . we have also demonstrated the capability of hfb gogny - d1s calculations to reproduce those features and therefore to make predictions in unexplored regions . this work was supported by micinn ( spain ) under research grants fis200801301 , fpa2009 - 08958 , and fis2009 - 07277 , as well as by consolider - ingenio 2010 programs cpan csd2007 - 00042 and multidark csd2009 - 00064 . one of us ( r.r . ) would like to thank both prof . j. ysto and dr . i. moore from the university of jyvskyl for valuable discussions . casten and b.m . sherrill , prog . 45 ( 2000 ) s171 ; m.v . c 68 ( 2003 ) 054312 ; d. vretenar _ et al . 409 ( 2005 ) 101 ; j. dobaczewski _ et al . 59 ( 2007 ) 432 .	the evolution of the ground - state nuclear shapes in neutron - rich sr , zr , and mo isotopes , including both even - even and odd-@xmath0 nuclei , is studied within a self - consistent mean - field approximation based on the d1s gogny interaction . neutron separation energies and charge radii are calculated and compared with available data . a correlation between a shape transition and a discontinuity in those observables is found microscopically . while in sr and zr isotopes the steep behavior observed in the isotopic dependence of the charge radii is a consequence of a sharp prolate - oblate transition , the smooth behavior found in mo isotopes has its origin in an emergent region of triaxiality . 21.60.jz , 21.10.ft , 27.60.+j
spin - orbit splitting of the energy dispersion relations for electrons in nonsymmetric quantum heterostructures has been theoretically considered during the past decades ( see ref . for a review ) . in bulk materials spin - orbit interaction appears both due to a slow - variable potential ( related to the lattice constant ) @xcite and due to cubic 3 and linear @xcite spin - dependent contributions to the effective hamiltonian . turning to the two - dimensional ( 2d ) case , we can reduce the cubic contribution to a linear one after the replacement of the squared momentum by the quantized value due to confinement @xcite . it is still more important the fact that we have to take into account an additional spin - orbit splitting of the energy spectrum due to the interaction with abrupt heterojunction potentials ( see ref . and discussion in refs . ) . such contribution is of a radically different kind with respect to the listed above because contributions from both sides of a slow confinement potential compensate each other @xcite and , therefore , the spin - splitting of 2d states can not be obtained without a short - range potential contribution . to the best of our knowledge , the relative contributions from the bulk - induced mechanisms @xcite and from the heterojunctions have not been clarified experimentally in spite of a set of existing theoretical calculations @xcite . in this paper we have examined the effect of an in - plane magnetic field on the electron energy spectrum and density of states and we have found that magneto - induced modifications of these characteristics are essentially different from the 2d spin - orbit interaction and zeeman splitting . in principle , this fact makes possible an experimental verification of the above discussed contributions of the spin - orbit interaction . ) and quantum well ( @xmath0 ) . dashed lines show electron ground state energy levels and thin curves correspond to wave functions ; @xmath1 is the qw s width , @xmath2 are the band offsets , and @xmath3 is the effective mass in the narrow - gap region , while @xmath4 and @xmath5 are the barrier effective masses . ] the effect of in - plane magnetic field on the energy spectrum in non - symmetric heterostructures occurs due to the zeeman term in the electron hamiltonian . this fact has been found in ref . , shortly after initial considerations of the spin - orbit splitting in non - symmetric heterostructures ( see refs . ) . a number of peculiarities in transport phenomena for 2d systems under an in - plane magnetic field were also discussed @xcite and more complicated cases , such as quasi-1d transport or spin hall effect ( see refs . in @xcite or @xcite , respectively ) were recently considered . all this papers only take into account the mix between the zeeman contribution and the effective 2d spin - orbit interaction . here we perform the calculations based on the three - band kane model@xcite with non - symmetric boundary conditions , which is valid for narrow - gap heterostructures , and with a slow potential described self - consistently ( see typical band diagrams in fig . the analysis of magneto - induced modifications of 2d energy spectra and corresponding density of states is presented for typical parameters of ingaas / inalas - selectively - doped heterojunctions , and ingaas and insb - based quantum wells ( qws ) . the paper is organized as follows . in the next section we discuss the conduction @xmath6-band eigenstate problem and describe the electronic states in non - symmetric narrow - gap heterostructures using the boundary conditions for the wave functions at the interfaces in the parabolic approximation . in sec . iii we solve the 2d - eigenstate problem using the averaged transverse field approach . numerical self - consistent calculations of the magneto - induced ( caused by a magnetic field ) modifications of the energy spectra and the density of states are presented in sec . conclusions are given in the last section . we start from the formulation of the eigenstate problem for the electronic state of the conduction @xmath6-band localized in the selectively - doped heterojunction or in the non - symmetric qw . all the above listed spin - dependent contributions are taken into account in the analysis . we also present the density of states which is connected to the photoluminescence excitation ( ple ) intensity . the electronic states in narrow - gap heterostructures are described by the three - band kane matrix hamiltonian @xcite @xmath7where the kinetic momentum , @xmath8 , contains the vector potential @xmath9 , @xmath10 is an in - plane magnetic field and @xmath11 is written in the @xmath12-representation through the 2d momentum @xmath13 . here we have also introduced the diagonal energy matrix @xmath14 whose elements determine the positions of the band extrema ( the energy values @xmath15 , @xmath16 , and @xmath17 correspond to electron , heavy- and light - hole extrema ) and the interband velocity matrix @xmath18 . the @xmath19 hermitian matrix @xmath20 is determined by the following non - zero matrix elements @xcite : @xmath21and @xmath22 is the characteristic interband velocity for the kane model . in such definitions eq . ( 3 ) may be rewritten as the spinor eigenvalue problem : @xmath36 _ { z}+\frac{g_{z}}{2}\mu _ { \scriptscriptstyle b}h\hat{\sigma}_{y}\right\ } \psi _ { \mathbf{p}z } & = & 0,\end{aligned}\]]where spinor @xmath37 is determined by the components @xmath38 , @xmath39 is bohr magneton , and @xmath40 is the pauli matrix . the energy values of @xmath6- and @xmath23-band extrema for the narrow - gap ( @xmath41 ) and wide - gap ( @xmath42 ) regions take the forms : @xmath43and the corresponding band diagrams for qw and selectively - doped heterojunction were shown in fig.1 . here the energy is counted from the bottom of @xmath6-band , @xmath44 is the gap , @xmath45 and @xmath46 are the band offsets for @xmath6- and @xmath23-bands , correspondingly . here the slow potential @xmath47 should be determined from a self - consistent procedure . thus , after substitution of ( 6 ) in eq . ( 5 ) we have formulated the eigenstate problem for the spinor @xmath37 . since @xmath48 in narrow - gap heterostructures , then the weak underbarrier penetration of wave function takes place for the electronic states with @xmath49 ( parabolic band approximation ) . because of this , one can neglect the longitudinal motion for the underbarrier region , @xmath50 for the single heterojunction case shown in fig.1@xmath51 , where the solutions take the form : @xmath52here @xmath53 and @xmath54determines the scale of the underbarrier penetration of wave functions . we have also used in eq . ( 7 ) the continuity conditions for eigenfunctions @xmath55 at heterojunctions . since @xmath32 in eq . ( 4 ) is proportional to @xmath56 , then the integration ( 5 ) over heterojunction produces additional contributions to the boundary condition for flows . such contributions are proportional to the band offset at the heterojunction @xmath46 . eliminating the underbarrier contributions from such equation by the use of the explicit expression ( 7 ) we obtain the third kind boundary conditions @xcite : @xmath57 _ { z}\psi _ { % \mathbf{p}z=0}=0.\]]here the momenta @xmath58 characterizes the underbarrier penetration and the parameter @xmath59 determines the spin - orbit coupling due to the abrupt potential of the heterojunction @xmath60and the right - side part is written for the approximation @xmath61 in the parabolic approximation , we describe the @xmath6-band electronic states using the above introduced spinor @xmath37 . below we consider not very strong magnetic fields ( @xmath62 , where @xmath63 is the fermi velocity and @xmath64 is the cyclotron frequency ) when @xmath65 in eq . ( 5 ) is replaced by @xmath66 and the isotropic kinetic energy is given by @xmath67 and includes the effective mass @xmath3 . considering the low - doped structure case ( if @xmath68 ) we rewrite the schrdinger equation ( 5 ) for the narrow - gap region in the form : @xmath69 _ { z}+\frac{g_{z}}{2}\mu _ { \scriptscriptstyle b}h\hat{\sigma% } _ { y}\end{aligned}\]]where @xmath47 is the self - consistent potential , as given in eq . ( 14 ) below , and the effective mass is @xmath70-independent . the spin velocity @xmath32 and the @xmath70-dependent contribution to the @xmath33-factor are proportional to the transverse electric field @xmath71 : @xmath72for the selectively - doped heterojunction case we consider eq . ( 10 ) for the @xmath73 region with the boundary conditions ( 8) and with @xmath74 . for the case of a qw of width @xmath1 , one can eliminate the underbarrier contributions in analogy to eqs . ( 7 - 9 ) and , in addition to eq . ( 10 ) for the region @xmath75 , the following third kind boundary conditions should be used : @xmath76 _ { z}\psi _ { \mathbf{p},\pm \frac{d}{2}}=0.\]]here momenta @xmath77 determine the scale of the underbarrier penetration of wave functions ; the different values of @xmath78 and @xmath5 take into account the differences of band offsets and effective masses , as it is shown in fig . 1@xmath79 . the parameters @xmath80 determine the spin - orbit coupling due to the abrupt potential of the heterojunction according to eq . when @xmath81 , which corresponds to the symmetric case , we will take @xmath82 and @xmath83 , as we can see from ( 9 ) . thus , both interface potentials , which determine the spin - dependent contributions to ( 12 ) , and the intrawell field , which determines the spin velocity @xmath32 in ( 11 ) , are responsible for the spin - splitting of energy spectra . since the zeeman spin - splitting term only appears in the eq . ( 10 ) , the mentioned magneto - induced modifications of electron states due to these two contributions are different . the self - consistent numerical procedure for the eigenstate problem ( 10 ) involves the potential @xmath47 , which is obtained from the poisson equation in the following form : @xmath84\]]here @xmath85 is the 3d concentration of donors and @xmath86 is the dielectric permittivity that we have supposed as uniform across the heterostructure . the electron density distribution is introduced through the electron dispersion relations @xmath87 according to : @xmath88where @xmath89 denotes the hermitian conjugate of @xmath90 and @xmath91 refers to the two possible spin orientations . the heaviside function , @xmath92 , appears here for the zero - temperature case . the fermi energy , @xmath93 , is expressed through the total electron density , @xmath94 , defined as @xmath95 thus , @xmath96 depends on @xmath97 for the fixed concentration case . the density of states is given by the standard formula : @xmath98 in order to analyze @xmath99 one needs to solve the above - formulated eigenstate problem and to perform the integrations in eq . the density of states is connected to ple intensity for the case of near - edge transitions , @xmath100 . since the interband matrix element @xmath101 do not depend on the in - plane quantum numbers , one obtains @xcite : @xmath102where @xmath103 is the polarization vector , @xmath104 and @xmath105 is the gap energy , which is renormalized due to the confinement effect before the numerical consideration we perform a simplified calculation of the eigenstate problem using the uniform transverse field approximation @xmath106 which is valid for non - doped qws under an external modulating field @xmath107 ; for heavy - doped structures @xmath107 implies an averaged self - consistent transverse field . the two @xmath70-dependent fundamental solutions @xmath119 ( labeled below by @xmath120 ) are determined from the equation @xmath121written for the narrow - gap region . the general spinor solution @xmath37 is expressed through these fundamental solutions according to @xmath122where the coefficients @xmath123 are determined from the boundary conditions ( 12 ) or from ( 8) and the requirements @xmath74 . after the substitution of the solution ( 21 ) for @xmath124 , which corresponds to the boundary condition at @xmath125 , into ( 8) and the multiplication of this system by @xmath112 on the left side we rewrite the boundary condition at @xmath126 as follows : @xmath127 \psi _ { \mathbf{p}z=0 } & = & 0 , \nonumber \\ \psi _ { \mathbf{p}}^{\scriptscriptstyle(-1)+ } { } \cdot \left [ \begin{array}{ll } 0 & -p_{+ } \\ p_{- } & 0% \end{array}% \right ] \psi _ { \mathbf{p}z=0 } & = & 0,\end{aligned}\]]with @xmath128 . in order to calculate the proportional to @xmath59 contributions we use here the solutions of the spin - dependent eigenstate problem ( 19 ) . thus , eq . ( 22 ) can be transformed with the use of the relations @xmath129 \psi _ { \mathbf{p}}^{(\sigma ) } & = & \sigma \frac{\overline{\mathrm{v}}% p^{2}+w_{\scriptscriptstyle h}p_{x}}{iw_{\mathbf{p } } } \nonumber \\ \psi _ { \mathbf{p}}^{(\sigma ) } { } ^{\scriptscriptstyle + } \cdot \left [ \begin{array}{ll } 0 & -p_{+ } \\ p_{- } & 0% \end{array}% \right ] \psi _ { \mathbf{p}}^{(-\sigma ) } & = & -i\frac{w_{\scriptscriptstyle % h}p_{y}}{\mathrm{v}p_{\sigma } + \sigma w_{\scriptscriptstyle h}}\end{aligned}\]]to a simple linear system for @xmath130 . the dispersion relation , @xmath131 , is determined from the zero determinant requirement . a similar transformation of the boundary conditions ( 12 ) for qws , after substitution on ( 21 ) , permit us to rewrite the boundary condition at @xmath132 in the form : @xmath133 \psi _ { \mathbf{p},\pm \frac{d}{2 } } & = & 0 , \nonumber \\ & & \left ( \hat{p}_{z}\mp ip_{\pm } \right ) \left ( a_{-}\varphi _ { z}^{% \scriptscriptstyle(a,-1)}+b_{-}\varphi _ { z}^{\scriptscriptstyle% ( b,-1)}\right ) \left\vert _ { z=\pm \frac{d}{2}}\right . \nonumber \\ + \chi \psi _ { \mathbf{p}}^{\scriptscriptstyle(-1)+ } { } \cdot \left [ \begin{array}{ll } 0 & -p_{+ } \\ p_{- } & 0% \end{array}% \right ] \psi _ { \mathbf{p},\pm \frac{d}{2 } } & = & 0.\end{aligned}\]]thus , we have obtained a linear system for @xmath130 and @xmath134 and , therefore , @xmath87 is obtained from the solvability condition for this system . first , let us consider the case of a heterostructure without spin - orbit contributions from heterojunctions , @xmath135 , when the dispersion relation is given by eq . for the zero magnetic field case , @xmath136 , the energy @xmath137 is replaced by @xmath138 and @xmath139 in eq . ( 19 ) is transformed to the isotropic dispersion relation @xmath140 . for this case the density of states was considered in ref . . if @xmath141 , the dispersion relation ( 19 ) becomes anisotropic as it is shown in fig . 2 @xmath51 . we have used in this figure the dimensionless magnetic field @xmath142 . in order to represent a general case , valid for any structure , we have also used dimensionless energy axis . values . solid , dashed and dot - dashed curves corresponds to the dimensionless momenta @xmath1430 , 0.25 , and 1 , respectively . ( b ) density of states vs dimensionless energy for magnetic fields @xmath1440 ( solid ) , 0.5 ( dashed ) , 1 ( dotted ) , 2.5 ( dash - dotted ) , and 5 ( dash - dot - dotted ) . , title="fig : " ] values . solid , dashed and dot - dashed curves corresponds to the dimensionless momenta @xmath1430 , 0.25 , and 1 , respectively . ( b ) density of states vs dimensionless energy for magnetic fields @xmath1440 ( solid ) , 0.5 ( dashed ) , 1 ( dotted ) , 2.5 ( dash - dotted ) , and 5 ( dash - dot - dotted ) . , title="fig : " ] for the case of the strong magnetic field , when @xmath137 is replaced by @xmath145 , the zeeman splitting effect appears to be dominant in the dispersion relation : @xmath146 . the spin - orbit splitting is mainly manifested as a shift of the dispersion paraboloids towards higher ( lower ) @xmath147 values for @xmath148 , respectively . on the other hand , zeeman splitting is shown as a displacement to higher ( lower ) energy values depending on @xmath149 . the anisotropy of the dispersion relation @xmath150 appears to be essential for the region around the cross - point , @xmath151 . after the shift @xmath152 and the integration over @xmath153 with the use of the @xmath154-function , the density of states is transformed from eq . ( 15 ) into the integral with respect to the cosine of the in - plane angle , @xmath155 : @xmath156here @xmath157 and @xmath158 . a straightforward integration for the above - cross - point region @xmath159 gives as the exact relation : @xmath160 , while for the below - cross - point region @xmath161 we plot @xmath162 versus dimensionless energy @xmath163 for different fields @xmath164 . figure 2@xmath165 shows this density of states for different dimensionless magnetic fields . two points deserve special attention . first , the typical @xmath166 singularity around @xmath167 , which occurs for @xmath168 , tends to disappears with increasing field . second , abrupt steps corresponding to the filling of each level are shifted with the field intensity indicating a delay in the filling caused by the zeeman splitting . it should be noted that the lowest @xmath169 level seems to be overfilled for low magnetic fields , i.e. , @xmath170 . electrons equally distributes between levels for magnetic fields higher than @xmath171 . here we perform numerical calculations based on the simplified consideration outlined in sec . ii@xmath42 , as well as the self - consistent solution of the eigenstate problem of sec . ii@xmath41 . the correspondent level - splitting at the fermi energy level in a selectively - doped heterojunction and the densities of states in qws are described . let us consider the 2d approach , when @xmath181 is around the ground energy level , @xmath182 , determined by the equation : @xmath183 . expanding @xmath184 $ ] over @xmath185 one can transform eq . ( 27 ) into a quadratic equation and the dispersion relation takes the form @xmath186 , where the splitting @xmath187 of the energy spectrum is given by @xmath188with the heterojunction - induced spin velocity , @xmath189 , and the internal spin velocity , @xmath190 , introduced in eq . ( 17 ) . in fig . 3 we plot these dispersion relations for the same parameters used in fig . 2@xmath51 . as in fig . 2@xmath51 , we have used dimensionless energy and momenta to represent a general case . comparing fig . 2@xmath191 with fig . 3 we can see the effect of the interface contribution as an enhancement of the splitting between parabolas @xmath91 . for different magnetic fields @xmath192 solid , dashed and dot - dashed curves corresponds to the dimensionless momenta @xmath1430 , 0.25 , and 1 , respectively . ] for high electron concentration the approach of eq . ( 28 ) is no longer valid . in order to obtain @xmath181 for this case we need to solve eq . ( 10 ) together with eq . ( 13 ) by means of a self - consistent approach . to do that we have used the transfer matrix method@xcite . figure 4 shows dispersion relations obtained in this way for in@xmath193al@xmath194as / in@xmath195ga@xmath196as with the electron density @xmath197 @xmath198 , which corresponds to a fermi energy @xmath199 mev . using the standard parameters @xcite one obtains @xmath200 and @xmath201 cm / s , so that dimensionless field @xmath202 corresponds to @xmath203 t. thus , we can compare panel for @xmath202 in fig.3 with panel for @xmath203 t in fig . 4 . we can see that numerically calculated results show a similar behavior than approximation ( 28 ) , although , for the material under consideration , splitting is qualitatively less and shift along momentum is larger in fig . 4 . and @xmath204 @xmath205 solid line : @xmath206 . dashed line : @xmath207 @xmath208 dot - dashed line : @xmath209 . ] the spin splitting of the dispersion relations at fermi energy is directly connected to the shubnikov - de haas ( sdh ) oscillations . since @xmath210 is nearly isotropic over @xmath13-plane , we have also calculated @xmath211 for different magnetic fields , see fig . 5 . following refs . the spin splitting is related to the modulation of the sdh oscillation amplitude according to @xmath212 . for the structure under consideration , experimental value@xcite is @xmath213 mev for @xmath214 t and @xmath197 @xmath198 , which corresponds to @xmath215 mev , calculated numerically for the same magnetic field and density . figure 5 shows a very slight dependence of @xmath216 on the magnetic field for @xmath200 being stronger this dependence for the case @xmath135 . thus , interface contributions should be detectable through the sdh oscillations amplitude . vs magnetic field for @xmath217 and @xmath218 . note that the upper @xmath216 is almost @xmath97-independent . ] ( solid line ) and @xmath229 ( dashed line ) . ( b ) density of states for @xmath26 @xmath230 ( solid line ) , @xmath231 @xmath230 ( dashed line ) , @xmath232 @xmath230 ( dot line ) , and @xmath233 @xmath230 ( dot - dashed line).,title="fig : " ] ( solid line ) and @xmath229 ( dashed line ) . ( b ) density of states for @xmath26 @xmath230 ( solid line ) , @xmath231 @xmath230 ( dashed line ) , @xmath232 @xmath230 ( dot line ) , and @xmath233 @xmath230 ( dot - dashed line).,title="fig : " ] once again , after the expansion near the ground energy , @xmath182 , one can transform the determinant in a similar way to transformations of eqs . ( 26 , 27 ) . numerical solutions of these equations for ingaas- and insb - based qws are represented in figs . 6@xmath191 and 7@xmath234 . we have considered @xmath235 wide ingaas- and insb - based qws using the parameters of refs . . we have also applied a field of @xmath236 @xmath237 , which corresponds to spin velocity @xmath238 cm / s , for the ingaas qw , and a field of @xmath235 @xmath237 , spin velocity@xmath239 cm / s , for the insb . comparing figs . 6@xmath191 and 7@xmath191 with fig . 4 we can see that , for similar spin velocities and magnetic fields , the splitting between levels is bigger whereas the shift in the @xmath147 direction is smaller for the qw case with respect to the selectively - doped structure . thus , the effect of interface contribution seems to be more pronounced for the qw case . this effect is enhanced as the width of the wells diminishes . due to the greater spin velocities of ingaas and insb - based qws , caused by the different effective masses and energy gaps , it is necessary the use of lesser magnetic fields to get a similar effect . both for qws and selectively doped structures , interface contributions are opposite to the intrinsic spin - orbit coupling effect . ( solid line ) and @xmath229 ( dashed line ) . ( b ) density of states for @xmath26 @xmath230 ( solid line ) , @xmath240 @xmath230 ( dashed line ) , @xmath231 @xmath230 ( dotted line ) , and @xmath232 @xmath230 ( dot - dashed line).,title="fig : " ] ( solid line ) and @xmath229 ( dashed line ) . ( b ) density of states for @xmath26 @xmath230 ( solid line ) , @xmath240 @xmath230 ( dashed line ) , @xmath231 @xmath230 ( dotted line ) , and @xmath232 @xmath230 ( dot - dashed line).,title="fig : " ] since the density of states is proportional to the ple intensity@xcite , as shown in eq . ( 16 ) , it is interesting to study the shape of @xmath241 , shown in figs . 6@xmath242 and 7@xmath242 . the effect of the interfaces is manifested in @xmath99 as a delay in the quenching of the @xmath243-singularity corresponding to the zero - field case . although the singularity no longer exists for @xmath244 , a peak still remains at the energy value of the bands anticrossing . for the case which includes interface contributions , this peak gets wider and @xmath99 tends slower to @xmath245 for high energy limit . thus , the ple technique is of great interest to analyze the interface contributions@xcite . it should be notice that @xmath91 levels becomes equally filled for magnetic fields beyond 5 t. in this paper we have examined the electron states in narrow - gap non - symmetric heterostructures under an in - plane magnetic field . the eigenstate problem was formulated in the framework of the three - band kane model with non - symmetric boundary conditions . we have found that the mechanisms of mixing between the zeeman term in the hamiltonian and the two kinds of spin - orbit coupling contributions ( from a slow field and from heterojunctions ) are essentially different . numerical estimates for typical parameters of ingaas / inalas and insb / inalsb structures demonstrate the essential magneto - induced modifications of the energy spectra under magnetic field strength of the order of tesla . let us discuss some possibilities for the experimental verification of the energy spectra modifications obtained here . the magnetotransport measurements of shubnikov - de haas oscillations under nearly in - plane magnetic fields@xcite ( when a quasi - classic quantization of the dispersion relations for the transverse component of the magnetic field is possible ) provide a direct information about in - plane magneto - induced modifications of electron energy spectra . the results of sec . iv@xmath41 are in agreement with ref . but more measurements ( for different in - plane fields and concentrations ) are necessary in order to separate the intrinsic and junction - induced contributions . another way of looking for these peculiarities of energy spectra is the mid - infrared ple spectroscopy@xcite when interband transitions are modified under in - plane magnetic field . ple intensity provides direct information of the energy spectra because it is directly connected to the density of states , as mentioned above . a comparison between sdh oscillations and ple measurements with a precision about 1 mev ( it should be possible for a high - quality structure ) provides more data on mechanisms of spin - orbit interaction in narrow - gap structures . thus , we have shown the essential modifications of the electron energy spectra in non - symmetric narrow - gap qws under in - plane magnetic fields . we suggest that measurements of the magneto - induced contributions to optical and transport properties would be an useful method for an experimental verification of the heterojunction - induced contributions to the spin - orbit interaction in the narrow - gap heterostructures under investigation . a. darr , j.p . kotthaus and t. ando , _ proc . f.g . fumi ( north - holland , amsterdam , 1976 ) , p. 774 ; t. ando , a.b . fowler and f. stern , rev . mod * 54 , * 437 ( 1982 ) ; d.a . romanov , phys . solid state * 35 , * 717 ( 1993 ) . j. luo , h. munekata , f. f. fang , and p. j. stiles , phys . b * 38 * , 10142 ( 1988 ) ; b. das , d.c . miller , s. datta , r. reifenberger , w.p . hong , p.k . bhattacharya , j. singh , and m. jaffe , phys . b * 39 * , 1411 ( 1989 ) . the final schrdinger eq . ( 10 ) and corresponding boundary conditions ( 8 , 12 ) do not change under taking into account the stress - induced splitting of @xmath23-band extrema and such a revision of the parameters changes the numerical estimation only slightly but the main results are the same . following refs . 11 - 13 and i. vurgaftman , j.r . meyer , and l.r . ram - mohan , j. appl . phys . * 89 * , 5815 ( 2001 ) , we use the data for in@xmath246al@xmath194as / in@xmath195ga@xmath196as structure : @xmath247 , @xmath248 , where @xmath249 is the electron mass , factor @xmath250 , permittivity @xmath251 , @xmath252 mev , @xmath253 mev , and @xmath254 mev . national compound semiconductor roadmap : insb , http://www.onr.navy.mil/sci_tech/information ( 2005 ) . data for insb / in@xmath255al@xmath194sb are @xmath256 , @xmath257 , @xmath258 mev , @xmath259 mev , @xmath260 mev , @xmath261 , @xmath262 .	modifications of spin - splitting dispersion relations and density of states for electrons in non - symmetric heterostructures under in - plane magnetic field are studied within the envelope function formalism . spin - orbit interactions , caused by both a slow potential and the heterojunction potentials ( which are described by the boundary conditions ) are taken into account . the interplay between these contributions and the magnetic field contribution to the spin - splitting term in the hamiltonian is essential when energy amount resulting from the zeeman and spin - orbit coupling are of the same order . such modifications of the energy spectra allow us to separate the spin - orbit splitting contributions due to a slow potential and due to the heterojunctions . numerical estimates for selectively - doped heterojunction and quantum well with narrow - gap region of electron localization are performed .
the common belief that the cmb is gaussian distributed can be directly traced back to the generic assumption that the quantum fluctuations of the inflaton field are placed in the vacuum state @xcite . relaxing this assumption might lead to detectable signatures in various astrophysical tests , most interestingly in future cmb and large - scale structure observations . in this note , we study cmb non - gaussian signatures predicted within inflationary models with non - vacuum initial states for cosmological perturbations . the model incorporates a privileged scale , which implies the existence of a feature in the primordial power spectrum . the model predicts a vanishing three - point correlation function for the cmb temperature anisotropies @xcite . we here focus on the first non - vanishing moment , the cmb four - point function at zero lag , namely the kurtosis , and compute its expected value for different locations of the primordial feature in the spectrum , as suggested in the literature to conform to observations of large scale structure @xcite . we consider non - vacuum states for the cosmological perturbations of quantum mechanical origin . let @xmath0 be a domain in momentum space , such that if @xmath1 is between @xmath2 and @xmath3 , the domain @xmath0 is filled by @xmath4 quanta , while otherwise @xmath5 contains nothing . the state @xmath6 is defined by @xmath7 the state @xmath8 is an @xmath4-particle state satisfying , at conformal time @xmath9 : @xmath10 and @xmath11 . we have the following property @xmath12 it is clear from the definition of the state @xmath13 that the transition between the empty and the filled modes is sharp . in order to `` smooth out '' the state @xmath13 , we consider a state @xmath14 as a quantum superposition of @xmath15 . in doing so , we introduce an , a priori , arbitrary function @xmath16 of @xmath17 . the definition of the state @xmath18 @xmath19 where @xmath16 is a given function which defines the privileged scale @xmath20 . we assume that the state is normalized and therefore @xmath21 . in the state @xmath22 , for any domain @xmath5 one has @xmath23 in these formulas , @xmath24 is a function that is equal to 1 if @xmath25 and 0 otherwise . these relations will be employed in the sequel for the computation of the cmb temperature anisotropies for the different non - vacuum initial states . the spherical harmonic expansion of the cosmic microwave background temperature anisotropy , as a function of angular position , is given by @xmath26 as we are interested in a non - gaussian signature of primordial origin we will be focusing on large angular scales , for which the main contribution to the temperature anisotropy is given by the sachs - wolfe effect , namely , @xmath27 , $ ] where @xmath28 is the bardeen potential , while @xmath29 and @xmath30 denote respectively the conformal times now and at the last scattering surface . note that the previous expression is only valid for the standard cold dark matter model ( scdm ) . in general , we might also be interested in the case where a cosmological constant is present ( @xmath31cdm model ) since this seems to be favored by recent observations . then , the integrated sachs - wolfe effect plays a non - negligible role on large scales and the expression giving the temperature fluctuations is not as simple as the previous one . in the theory of cosmological perturbations of quantum mechanical origin , the bardeen variable becomes an operator , and its expression can be written as @xmath32~,\ ] ] where @xmath33 is the planck length . in the following , we will consider the class of models of power - law inflation since the power spectrum of the fluctuations is then explicitly known . in this case , the scale factor reads @xmath34 , where @xmath35 is a priori a free parameter . however , in order to obtain an almost scale - invariant spectrum , @xmath36 should be close to @xmath37 . in the previous expression of the scale factor , the quantity @xmath38 has the dimension of a length and is equal to the hubble radius during inflation if @xmath39 . the parameter @xmath38 also appears in eq . ( [ bop ] ) . the mode function @xmath40 of the bardeen operator is related to the mode function @xmath41 of the perturbed inflaton through the perturbed einstein equations . in the case of power - law inflation and in the long wavelength limit , the function @xmath42 is given in terms of the amplitude @xmath43 and the spectral index @xmath44 of the induced density perturbations by @xmath45 using the rayleigh equation and the completeness relation for the spherical harmonics and after some algebra we get @xmath46 j_{\ell } [ k(\eta _ 0-\eta _ { \rm lss})]y_{\ell m}^*({\bf k})~. \label{alm}\ ] ] at this point we need to somehow restrict the shape of the domain @xmath5 . we assume that the domain only restricts the modulus of the vectors , while it does not act on their direction . then , from eq . ( [ alm ] ) , one deduces @xmath47 \delta _ { \ell _ 1\ell_2}\delta _ { m_1m_2}~,\end{aligned}\ ] ] @xmath48 k^3\|f_k\|^2\frac{{\rm d}k}{k}\ ] ] thus , the multipole moments @xmath49 , in the state @xmath13 , are given by @xmath50 where @xmath51 is the `` standard '' angular power spectrum , i.e. , the multipole obtained in the case where the quantum state is the vacuum , i.e. , @xmath52 . let us calculate the same quantity in the state @xmath53 . performing a similar analysis as the above one , we find @xcite @xmath54\delta _ { \ell _ 1\ell_2}\delta _ { m_1m_2}~,\end{aligned}\ ] ] @xmath55 where , to reach this eqn , we defined @xmath56 [ we will see below that this function @xmath57 can not be arbitrary ] and we integrated by parts , and then we defined @xmath58 $ ] . in this we have not assumed anything on @xmath59 or @xmath60 . we see that the relation @xmath61 requires the function @xmath62 to be monotonically increasing with @xmath63 . it is interesting that , already at this stage of the calculations , very stringent conditions are required on the function @xmath62 which is therefore _ not _ arbitrary . this implies that the function @xmath64 which appears in the correction to the multipole moments is always positive , vanishes at infinity and is monotonically decreasing with @xmath63 . an explicit profile for @xmath64 is given in fig . the total power spectrum of the bardeen potential can be written as @xmath65\biggr\}=a_{_{\rm s}}k^{n_{_{\rm s}}-1 } [ 1 + 2n\bar{h}(k)].\ ] ] observations indicate that @xmath66 and for simplicity we will take @xmath67 . as we have seen previously , we can write the multipole moments in the state @xmath14 as @xmath68 . substituting the well - known expression for the @xmath51 s and the definition of @xmath69 given by eq . ( [ defdl2 ] ) , one finds that the coefficients @xmath70 are given by @xmath71 \over \gamma^2[(4-n_{\rm s})/2]\gamma[\ell -(n_{\rm s}-5)/2 ] } + 2n{\bar d_{\ell } } ^{(2)}\biggr\}.\ ] ] as a next step , one has to normalize the spectrum ( need to determine the value of @xmath72 ) . we choose to use the value of @xmath73^{1/2 } ( \ell_{\rm pl}/\ell_0 ) \sim 18 \mu k$ ] with @xmath74 measured by the cobe satellite . thus , we compute the quadrupole and then @xmath75^{-1},\ ] ] for @xmath67 . the band power @xmath76 gives @xmath77 the @xmath4-dependence in the above expression is the correction due to the non - vacuum initial state . we easily check that if @xmath52 the corresponding band powers are constant at large angular scales . finally , we calculate the two - point correlation function at zero lag in the state @xmath78 . using eqs . ( [ dtt ] ) , ( [ clpsi2final ] ) , the second moment , @xmath79 , of the distribution is given by @xmath80 ^ 2\biggr \rangle = \frac{\ell_{\rm pl}^2}{\ell_0 ^ 2 } \sum _ { \ell } \frac{2\ell + 1}{4\pi } c_{\ell } ^{(2 ) } ~.\ ] ] once we have reached this point , an obvious first thing to do is to check that the two - point correlation function calculated above is consistent with present observations . among the available observations that one can use to check the predictions of theoretical models , two are key in cosmology : the cmb anisotropy and the matter - density power spectra . we will not study in details all the predictions that can be done from the two - point correlation function since our main purpose in this note is to calculate the non - gaussianity which is a clear specific signature of a non vacuum state . so , we just compute the matter power spectrum to demonstrate that it fits reasonably well the available astrophysical observations for some values of the free parameters . in addition , this illustrates well the fact that , using the available observations , we can already put some constraints on the free parameters . a simple ansatz for the function @xmath64 is represented in fig . [ fig1 ] and can be expressed as @xmath81~.\ ] ] with it , the matter power spectrum today , after taking into account the transfer function @xmath82 which describes the evolution of the fourier modes inside the horizon , can be written @xmath83^{-1 } \biggl[1 + 2n\bar{h}(k)\biggr]k_{\rm phys}~.\ ] ] the scdm transfer function is given approximatively by the numerical fit @xcite @xmath84^{-1/4},\ ] ] where @xmath85 and the shape parameter @xmath86 can be written as @xmath87 \quad \quad \gamma \equiv \omega _ 0 h e^{-\omega _ { \rm b}- \sqrt{2h}\omega _ { \rm b}/\omega _ 0 } ~,\ ] ] where @xmath88 is the total energy density to critical energy density ratio and @xmath89 represents the baryon contribution . or , more explicitly , we take @xmath90 . we have now normalized the matter power spectrum to cobe . it is important to realize that the above procedure only works for the scdm model since we have used the sachs - wolfe equation . the scdm matter power spectrum is depicted in fig . [ fig1 ] . the measured power spectrum of the _ iras _ point source catalogue redshift survey ( pscz ) @xcite has also been displayed for comparison . one notices that the effect of the step in @xmath64 is to reduce the power at small scales which precisely improves the agreement between the theoretical curves for @xmath91 and the data . let us remind at this point that the shape of the function @xmath64 has not been designed for this purpose and comes from different ( theoretical ) reasons . therefore , it is quite interesting to see that the power spectrum obtained from our ansatz fits reasonably well the data . this plot also confirms the result of ref . @xcite , namely that the number of quanta must be such that ( for scdm ) @xmath92 , i.e. , it can not be too large . it is time to proceed with the calculation of the four - point correlation function . briefly , one has to first perform the calculation for the state @xmath93 , and then generalize it for @xmath94 . after some lengthy but straightforward algebra , one finds @xcite @xmath95 \delta _ { \ell _ 1\ell _ 3}\delta _ { \ell _ 2\ell _ 4 } \delta _ { m_1,-m_3}\delta _ { m_2,-m_4 } \nonumber \\ & & + ( -1)^{m_1+m_2}\biggl[c_{\ell _ 1}c_{\ell _ 2 } + 2nc_{\ell _ 1}d_{\ell _ 2}^{(2 ) } + 2nc_{\ell _ 2}d_{\ell _ 1}^{(2 ) } + 4n^2f_{\ell _ 1 \ell _ 2}^{(2)}\biggr ] \delta _ { \ell _ 1\ell _ 4}\delta _ { \ell _ 2\ell _ 3 } \delta _ { m_1,-m_4}\delta _ { m_2,-m_3 } \nonumber \\ & & + ( -1)^{m_1+m_3}\biggl[c_{\ell _ 1}c_{\ell _ 3 } + 2nc_{\ell _ 1}d_{\ell _ 3}^{(2 ) } + 2nc_{\ell _ 3}d_{\ell _ 1}^{(2 ) } + 4n^2f_{\ell _ 1 \ell _ 3}^{(2)}\biggr ] \delta _ { \ell _ 1\ell _ 2}\delta _ { \ell _ 3\ell _ 4 } \delta _ { m_1,-m_2}\delta _ { m_3,-m_4 } \nonumber \\ & & -2n(n+1)e_{\ell _ 1\ell _ 2\ell _ 3\ell _ 4}^{(2 ) } { \cal h}_{\ell _ 1\ell _ 2\ell _ 3\ell _ 4}^{m_1m_2m_3m_4 } e^{i\pi(\ell _ 1+\ell _ 2+\ell _ 3+\ell _ 4)/2 } \biggl[1+(-1)^{\ell _ 1+\ell _ 3 } + ( -1)^{\ell _ 2+\ell _ 3}\biggr]\biggr \}~,\end{aligned}\ ] ] @xmath96 \\ \label{def4bb } e_{\ell _ 1\ell _ 2\ell _ 3\ell _ 4}^{(2 ) } & \equiv & \int _ 0^{+\infty } j_{\ell _ 1}[k(\eta _ 0-\eta_{\rm lss } ) ] \ldots j_{\ell _ 4}[k(\eta _ 0-\eta _ { \rm lss } ) ] \bar{h}(k)k^3\|f_k\|^4\frac{{\rm d}k}{k } \end{aligned}\ ] ] we are now in a position to calculate the cmb excess kurtosis . in order to establish an analytical formula for it , one just needs to use the equation linking @xmath97 and @xmath98 and play with the properties of the spherical harmonics . explicitly , the excess kurtosis is defined as @xmath99 where the second moment has already been introduced and where the fourth moment , @xmath100 , of the distribution is defined as @xmath101 with @xmath102 ^ 4 $ ] . an important shortcoming of the previous definition is that the value of @xmath103 depends on the normalization . it is much more convenient to work with a normalized ( dimensionless ) quantity . therefore , we also define the normalized excess kurtosis as @xmath104 which is the one more commonly used in the literature . we have performed the full numerical resolution of the normalized excess kurtosis parameter which we show in fig . [ fig2 ] . as we see from it , @xmath105 is an asymptotic value , provided we concentrate on the middle and big values of the built - in scale @xmath20 . an analytical order of magnitude estimate of this can be seen in ref . the fact that the numerical estimate does not depend on @xmath20 is confirmed by the plot , except for small values of the wave - numbers . in fact , this shows that the quantity @xmath106 does not depend very much on the free parameters . one can see this is true for parameters @xmath107 and @xmath4 since , using the above mentioned analytical estimate , we find , for @xmath108 , @xmath109 and for @xmath110 , @xmath111 . since we know that this result does not depend on the details of the weight function @xmath64 , we conclude that the asymptotic value obtained above is a generic value , at least for large values of @xmath20 . in particular , this is true for @xmath112 which corresponds to the built - in scale located roughly at the privileged scale in the matter power spectrum selected by the redshift surveys of ref . another important remark is that the excess kurtosis is found to be negative . the above paragraphs show that the introduction of non - vacuum initial states leads to a clear prediction for the excess kurtosis parameter . however , this by itself does not imply that the signature will be observable . the signal to theoretical noise ratio has to be considered or , stating it differently , we need to make the comparison of the signal against the cosmic variance . the cosmic variance quantifies the theoretical error coming from the fact that , in cosmology , observers have only access to one realization of the @xmath113 stochastic process whereas theoretical predictions are expressed through ensemble averages . the specific computation of the cosmic variance for the model under consideration was performed @xcite and found to be some four orders of magnitude higher than the signal itself . this is no surprise : we are interested in a non - gaussian signature of primordial origin , and thus we are focusing on large angular scales , for which the theoretical uncertainties are the highest . shifting to intermediate angular scales , a stronger signal would be obtained ; however , in that case secondary sources would be more difficult to subtract and thus the transparency of the effect would be compromised . in sum , the excess kurtosis is found to be negative and the signal to noise ratio for the dimensionless excess kurtosis parameter is equal to @xmath114 , almost independently of the free parameters of the model . this signature turns out to be undetectable and therefore we conclude that , subject to current tests , gaussianity is a generic property of single field inflationary models . i d like to thank my collaborators jrme martin and mairi sakellariadou for extensive discussions and for hospitality in paris . this work was partially financed with funds from conicet , uba , iap and fundacin antorchas .	in the framework of inflationary models with non - vacuum initial states for cosmological perturbations , we study non - gaussian signatures on the cosmic microwave background ( cmb ) radiation produced by a broken - scale - invariant model which incorporates a feature at a privileged scale in the primordial power spectrum .
this preprint continues preprints 1204.0200 and 1204.2423 and represents some results of numerical simulations of auto - generation magnetic - wave chaos in ferrite films , its synchronization and application to secure communication . a references like ( 6.x ) means formula ( x ) from section 6 placed in 1204.2423 . sections 1 - 5 are placed in 1204.0200 . our main interest below is \(i ) visual investigation of regular and chaotic non - linear magnetostatic spin wave patterns auto - generated through feedback consisting of wire inductors ( antennae ) , amplifier and may be filters ; ( ii ) investigation of those conditions of the auto - generation , and properties of generated patterns , what are mostly responsible for characteristics of resulting chaotic microwave - band electric signals ( voltages and currents ) , in particular , their possibilities to synchronize chaotic patterns in other similar systems ( ferrite film ) and thus to serve for secure transmission of information . 7.1 . chaotic auto - generation in film with linear feedback . if the voltage ( emf ) signal , @xmath0 , induced in a conductor ( antenna ) by magnetization precession is amplified and transformed into driving current , @xmath1 , in another conductor , then auto - generation of mw may take place . the simplest variant of this feedback is drawn at plot ( a ) in fig.13 . it consists of two identical loop inductors and purely linear amplifier whose gain and phase shift are with frequency independent . typical features of chaotic generation in such the scheme , at @xmath2 , are presented by figs.13 and 14 . the plots ( b ) and ( e ) in fig.13 show that rather wide - band magnetization chaos is produced , in frequency range about 600 mhz . position of most intensive peak at the voltage spectrum well corresponds to frequency of the damon - eshbach ( de ) surface wave with length @xmath3 @xmath4 dictated by the loop width , i.e. to @xmath5 @xmath6 ( @xmath7 ghz at@xmath8 @xmath9 ) . the lower edge of the spectrum lies below uniform precession frequency@xmath10 ( @xmath11 ghz @xmath8 ) . this means that the main - branch bulk waves also are excited . the upper edge of the voltage spectrum is far below the upper frequency of msw , @xmath12 ( @xmath13 @xmath14 ghz ) . hence , long surface msw are dominating in magnetization patterns . this pattern is seen at plot ( a ) in fig.14 . the contour plot of its spatial spectrum ( ( f ) in fig.14 ) demonstrates tracks of ( i ) a set of long - wave modes with frequencies about @xmath15 , ( ii ) three times shorter than @xmath16 de mode with frequency @xmath17 ( @xmath18 ghz ) , ( iii ) comparatively short - wave surface modes with non - zero transversal wavenumber , @xmath19 , and frequencies about @xmath20 , and , besides , ( iv ) approximately two times longer than @xmath16 de mode with frequency @xmath21 ( @xmath22 ghz ) . interestingly , the latter mode seems be responsible for lower high peak in the voltage spectrum . plots ( b ) and ( c ) show that , in rough terms , auto - generation switches between two states whose frequencies are about @xmath23 and @xmath24 , and that at the first of them the energy is usually lower than at the second ( in accordance with negative non - isochronity , see sec.5 ) . due to linearity of amplifier , the film itself are forced to take all cares about nonlinear saturation of precession angle . naturally , this results in complicated large - amplitude chaos characterized by wide variety of time scales and intermittency of different regimes , which is illustrated by plots ( b)-(d ) in fig.13 and ( b),(c ) and ( e ) in fig.14 . low - frequency contents of this chaos is better visible in the spectrum of energy , @xmath25 , which is enriching at least down to 5 mhz . voltage and magnetization time series yield fractal dimension @xmath26 . hence , we encounter hyper - chaos governed by not less than four relevant variables . 7.2 . auto - generator with non - linear feedback . as practice says , in real generators an inner non - linearity of amplifiers is essential . it manifests itself in two ways : ( i ) as non - linearity of input resistance which effectively saturates input voltage , and ( ii ) as saturation of output power or , equivalently , of output driving current . besides , one should take into account frequency dependence of gain , @xmath27 , and own impedances ( capacities and inductances ) of feedback conductors . corresponding scheme looks as in fig.15 . here @xmath28 , @xmath29and @xmath30 are linear ( small - amplitude ) input resistance , @xmath31 ohm , inductance of passive ( feedback ) antenna , @xmath32 nh , and capacity of active ( driving ) antenna , @xmath33 pf , respectively . since in reality such the circuit is quasi - stationary one ( its size is much smaller than lengths of em - waves under operation ) , a simple analysis shows that there is no necessity in additional reactive elements . for practical high - frequency wide - band amplifiers , typical level of the input saturation is @xmath34 @xmath35 v , level of output saturation is @xmath36 @xmath37 ma , while @xmath38 @xmath39 @xmath40 , and linear frequency characteristics can be modeled by @xmath41 with @xmath42 describing time delay . in numerical simulations , usually values @xmath43 ghz and @xmath44s@xmath45were used , as in ina-32063 wideband silicon rfic amplifier ( @xmath42 is small as compared with time delays in film and therefore insignificant ) . fig.15 shows how numerical copy of this scheme generates at relatively small gain , @xmath46 , and moderate output power saturation . we see that likely periodic ( although more or less complicated ) oscillations take place in all the amplitudes and phases . if decreasing gain and /or output saturation level one may observe more simple oscillations , down to trivial mono - chromatic regime and failure of generation , while increase results in further complications and then chaos . 7.3 . role of antennae separation . the fig.16 illustrates chaotic auto - generation in numerical model corresponding to @xmath47mm@xmath48mm@xmath49m - size film and @xmath50 mm width of loop inductors , or to a proportionally re - scaled system ( but not too small ) , at field @xmath51oe , moderate gain and relatively high output power saturation . at top of fig.16 static magnetization of such shaped film , at @xmath52 oe and @xmath53oe , is presented . clearly , a noticeable demagnetization takes place at film corners only . in this numerical experiment , two variants of separation , @xmath54 , of feedback and active inductors were considered , with @xmath55 @xmath56 @xmath57and @xmath58 @xmath59 @xmath60 , where @xmath61 mm is maximum of wavelengths preferably selected by the loop antennae . comparison of these cases demonstrates , at plot ( d ) , that auto - generation spectrum is sensitive not only to the @xmath62 but also to @xmath63 . it could be expected , because any phase shift in feedback should be compensated by equal magnetic wave phase difference , @xmath64 , between antennae , with @xmath65 being a dominating wave vector . at smooth feedback frequency characteristics , one can suppose that @xmath66 , therefore @xmath67 where @xmath68 is corresponding change in frequency and @xmath69 is group velocity . according to this estimate , transition from @xmath70to @xmath71 must result in increase of auto - generation frequencies , which agrees with numerical simulation . taking @xmath72 be the frequency of de wave , we obtain @xmath73 @xmath74 and then @xmath75 @xmath76 ( in dimensionless units ) , that is@xmath77 mhz , in satisfactory quantitative agreement with plot ( d ) . let us more discuss the voltage spectrum . as in case of linear feedback ( see above ) , its upper - frequency peak can be surely related to de mode with @xmath78 . but what is origin of the lower - frequency peak ? from watching for dynamics of magnetization pattern , it is possible to relate this peak to a long de mode with wavelength @xmath79 where @xmath80 . such mode also can get suitable phase conditions under feedback determined by eq.1 . hyper - chaos under non - linear feedback . non - linear concurrence between the two dominating wave modes manifests itself in chaotic jumps of instant ( time - local ) frequency of auto - generation , @xmath81 , as shown by plots ( a ) and ( c ) in fig.16 . it should be underlined that from the point of view of energy pattern ( quasi - local energy , see sec.6.14 ) the same interplay of different modes looks as the birth and drift of envelope solitons . in these terms , the frequency separation of two main peaks at plot ( d ) is nothing but mean number of soliton births per second ( see below ) . due to high power saturation level , the whole picture occurs very chaotic . its correlation dimension obtained from voltage and magnetization time series ( solid and dot lines , respectively ) is @xmath82 ( if not greater ) . 7.5 . weak chaos in generator with wire inductors . it is interesting how the picture will be changed if two - element loop inductors are replaced by wires , at the same feedback circuit . figs.17a and 17b can give the answer for case @xmath83 ( = 1.5 mm ) with @xmath63 being distance between the wires . in this case , two dominating modes chosen by the system are long de waves with lengths @xmath84 and @xmath85 . in spatial spectrum of magnetization ( in momentum space ) the mode with length @xmath86 can be detected , but practically it does not contribute to voltage spectrum . besides , as usually , a set of comparatively weak excited equal - frequency modes with @xmath87 is present . in real space non - linear magnetization pattern characteristic rhombic structuring is well visible invoked by dispersing properties of underlying linear mw eigenmodes . but frequencies of the latter become lowered by non - linearity . at nearly the same output saturation and even greater linear gain , the whole picture is significantly more smooth and less chaotic than in case of loop inductors , with @xmath88 only ! in part , the reason is that both the emf accepted by wire inductor and power pumped by it are approximately two times smaller than by loop inductor . in this concrete experiment , power absorption @xmath89 @xmath90 @xmath91 mw , where @xmath92 and @xmath93 are driving current and emf in active inductor , respectively , and @xmath94means time average ( over a few precession periods ) . 7.6 . generation of envelope solitons . in the middle plot in fig.17b , typical spatial distribution of the quasi - local energy density ( sec.6.14 ) is shown ( to be precise , of @xmath95 where @xmath96 is @xmath97units ) . it highlights presence of three envelope solitons of whose the first ( most left ) is in birth , the second is just passing through feedback inductor and the third is dying remainder of previously passed soliton . obvious rhombic shaping of the solitons reflects four most preferred directions of local energy propagation ( corresponding to maximum group velocity ) , @xmath98 @xmath99 @xmath100 ( see sec.5.4 ) . notice that solitons are very sharp ( as measured in @xmath54-direction ) : their width is even less than wavelengths of carry msw enveloped by them ! but several wavelengths have time to pass through such the envelope while it itself displaces on its width . 7.7 . instanton structuring of spin precession . let us go back to strong chaos considered above in subsection 4 and have a look how envelope solitons manifest themselves in local magnetization time series . the fig.18 shows behavior of spin and of @xmath101 , where @xmath102 is quasi - local energy density ( sec.6.14 ) , at a point close to the feedback inductor . notice that @xmath103 where @xmath104 is the wave function ( see in sec.5.3 and eq.5.9 there ) , and @xmath105 is the eccentricity ( see eqs.4.6 and 5.10 ) , i.e. can be interpreted as merely squared angle of precession . comparison of plots ( c ) and ( d ) in fig.18 visually prove that at fast time scale ( measured in periods of precession ) @xmath106is almost integral of motion . what most of all impresses in plots ( b ) , ( c ) and ( f ) is that @xmath107 ( and thus local angle of precession ) never overlooks to regularly turn into either exact zero or to extremely small value . if @xmath108 was mere random field ( like thermally activated magnetization noise in thermodynamical equilibrium ) such behavior would be absolutely improbable . it gives best evidence that we observe results of essentially nonlinear self - organization of magnetization field in `` clots '' . as in nonlinear field theory , spatially local time tracks of self - organization may be termed instantons . as it was mentioned in sec.5.4 , to be more precise , the eccentricity , @xmath105 , should be related to dominating msw mode ( instead of uniform one ) . in other words , the quasi - local energy introduced in eq.6.6 must be redefined with taking into account non - local ( gradient dependent ) part of energy estimated by eq.6.7 . corresponding refinement reads @xmath109 @xmath110 ^{1/2}\ ] ] where @xmath111is module of dominating wave - vector ( for loop inductors , @xmath112 is taken in mind . for comparatively short - wave chaos , this is better invariant of fast motion ( precession ) than what defined by ( 6.6 ) . .... 7.8 . phase slips and noise production . plots ( d ) and ( e ) in fig.18 show that when local amplitude of precession is passing through zero its phase can slip by some angle . in contrast to ideal black solitons ( see sec.5.6 ) , this angle can take any value different from @xmath113 . of course , usually two close spins undergo the same slip , but sometimes one or another spin can be beaten from common behavior . hardly this fine detail of magnetization dynamics is unambiguously determined by a few relevant chaotic variables . therefore , local phase slips may become the source of some amount of noise , in addition to chaos . although , by our observations , this noise is rather weak , probably it may create obstacles to ideal copying of chaos under synchronization . quasi - periodic chaos . fig.19 illustrates example of quasi - periodic chaos obtained in auto - generator with relatively narrow loop antennae ( @xmath114 mm ) at moderate gain ( = 9 ) and output saturation ( @xmath115 ma ) , and antennae separation @xmath116 . dominating wave - length of magnetization pattern is close to @xmath117 while carrier frequency of its oscillations is rather close to frequency of de wave with @xmath118 and @xmath119 . in this case , time series of different variables ( magnetic energy , magnetization and voltage envelopes and phases , instant frequency , etc . ) possess characteristic features of the kind of chaotic attractor termed noisy limit circle . the voltage spectrum at plot ( f ) resembles ruled spectra of complex periodic signals ( with period being inverse distance between neighboring spectral lines ) . nevertheless , in fact voltage signal is chaotic and its fractal dimension , @xmath120 ( see plot(g ) ) even indicates hyperchaos . plot ( d ) shows that in spite of large magnitude of instant frequency oscillations ( @xmath121 mhz ) and therefore very wide total frequency band of generated signal ( @xmath122ghz ) its phase modulation lies between well certain boundaries not exceeding @xmath123 . from comparison of plots ( d ) and ( e ) it is seen that positive ( negative ) slope of time - smoothed phase trajectory corresponds to smaller ( greater ) energy , in accordance with negative non - isochronity of mw . what here happens ? the frequency separation of the three main groups of lines in plot ( f ) , @xmath124mhz , is by its sense the frequency of soliton births near active inductor . hence , one envelope soliton is born approximately each @xmath125ns . the group velocity of de waves at operational frequency ( see eq.5.14 ) is @xmath126 cm / s , therefore time of soliton flight from active to passive loop is about @xmath127 ns . mean duration of one cycle of the quasi - periodic modulation can be estimated from plot ( e ) as @xmath128 ns . comparing these numbers we conclude that two solitons simultaneously are in action and that each cycle contains birth and transfer of @xmath129 solitons . what is for the cycle duration itself , seemingly it can be connected to the relaxation time , @xmath130 . role of film width to wave length ratio . intuitively , one may predict that formation and propagation of envelope solitons is sensitive to the ratio @xmath131where @xmath132is width of film and @xmath133 ( @xmath134 for two - element loop antennae ) @xmath8is dominating wave length in magnetization pattern . the fig.20a shows that , indeed , this is confirmed by numerical simulation of auto - generation in narrow ( in the sense that @xmath135 ) and wide ( @xmath136 ) films . in narrow film , soliton takes all width of a film . in wide film , at very beginning of next soliton manufacturing one energy clot may be inflated occupying all the width but soon it breaks into two or three ( or even more ) solitons each of which takes a part of the width only . then solitons run not in parallel to film boundaries but along one of two easy propagation directions which correspond to maximum group velocity ( see sec.5 ) . fig.20a demonstrates this for the case when two soliton chains are created ( of course , if three or five soliton chains realize then central of them moves in parallel to boundaries ) . importantly , in general in wide film ( at @xmath137 ) wave fronts ( equal - phase lines ) of magnetization pattern stronger are curved then in narrow film ( especially in region where different soliton chains meet one another ) . therefore , the wider is film the smaller is voltage ( emf ) per unit length induced in direct antenna , up to that wider film can produce lower voltage ( at the same precession angle ) which worsens the feedback . as example of unpretentious situation , fig.21 presents movie of one cycle of chaotic soliton generation in moderately pumped narrow film . 7.11 . about coherence of auto - generation . it is known that spectrum of any oscillation with arbitrary amplitude modulation and arbitrary ( let random ) but limited phase modulation consists of infinitely narrow line ( at carrier frequency ) surrounded by some pedestal . an unlimited phase modulation that is accumulation of phase slips only produces a broadening of the carrier line . in our numerical simulations of the auto - generation ( even concerning strongly chaotic regimes ) accumulation of phase slips ( phase diffusion ) on the average did not exceed one - two periods of carrier frequency per microsecond . hence , coherent generation was observed , to the extent of calculation duration ( apparent broadening of carrier line on spectrum plots is artifact of short - time fourier transform ) . theoretically , the degree of coherence of chaotic generation is interesting open question . from practical point of view , it may be better answered in real experiments which reflect realities ( like temperature instability ) not accounted for by numerical model . spatial differentiation of auto - generation spectrum . under auto - generation , magnetization in the whole film oscillates coherently . due to chaotic soliton production , local phases of common oscillation undergo chaotic deviations which however remain limited ( usually not greater then @xmath138 ) . local phase slips do not destroy common coherent picture . but , of course , details of pedestal of oscillation spectrum may strongly depend on what variable is under measuring . this statement is illustrated by fig.20a . the fig.20a corresponds to wide film , with @xmath139 and comparatively large separation of loop inductors , @xmath140 . it presents example of rather wideband chaos with complicated spectrum which in 6 peaks can be resolved . again , the distance between central ( most powerful ) carrier peak and its closest neighbors equals to inverse mean time separation of soliton births . approximately twice larger distance between the carrier and next two neighbors can be interpreted as second harmonics of the births frequency . naturally , the carrier peak is most brightly represented in spectrum of spatial fourier transform of magnetization at @xmath141(plot ( c ) ) and least brightly in spectrum of local spin precession ( plot ( b ) ) . circumferential synchronization of auto - generation . naturally , the question arises about possibility of synchronization of chaotic auto - oscillations in one generator by voltage signal taken from another ( at least identical ) generator . firstly , we tried to use voltage ( emf ) , @xmath142 , of additional wire inductor placed at right - hand end of the master film ( i.e. behind the feedback inductor ) . this voltage was compared with voltage , @xmath143 , produced in identical control inductor in the slave film . the difference , @xmath144 , was transformed into current , @xmath145 , passed through one more wire inductor close to the control one , as shown in fig.22 . notice that in case of exact coincidence , @xmath146 , the synchronizing current turns into zero , @xmath147 , and thus slave film feels no difference from master film . paradoxically , because of almost ideal coherence of chaotic auto - generation , at available duration of numerical experiments , high - frequency ( carrier ) synchronization is automatically executed and thus calls no problem . the problem consists of synchronization of phase and amplitude chaotic modulation . films with @xmath148 were investigated . various linear and non - linear ( saturated ) connections between @xmath144 and @xmath149 were tested , but none of variants has shown an evident superiority . typical results are illustrated by plots ( 3)-(8 ) in fig.22 which show three pieces of synchronization with respect to envelopes of feedback voltages in the master and slave , @xmath150 and @xmath151 ( see scheme in fig.15 ) , and to their instant frequencies . undoubted signs of partial synchronization can be noticed already in 100 ns after beginning of the process . they become better visible if take into account obvious delay of the slave response . plots ( 9 ) and ( 10 ) show how the picture looks if correction by delay is made . most of details of signal @xmath152 circulating in feedback device of master generator are reproduced by similar signal @xmath151 in the slave . according to plot ( 2 ) , high - frequency ( carrier ) phase difference between these signals does not leave the frame @xmath153 . nevertheless , in respect to the modulation there are essential quantitative and sometimes qualitative distortions . the results are far from copying of chaos . to estimate them we need in a more soft criterion for partial synchronization . defects of synchronization . careful analysis of data reflected in fig.22 yields that most part of misalignment between @xmath150 and @xmath151can be related to two reasons : ( i ) not constant but slowly time - varying delay , @xmath154 , and ( ii ) slowly time - varying amplitude shift which is in close correlation with difference between energies of master and slave film , @xmath155 @xmath156 @xmath157 . the latter statement is confirmed by plot ( 11 ) . notice that the delay , as shown at plot ( 12 ) , may be both positive and negative ( which corresponds to anticipating synchronization ) . this findings mean that , approximately , the relation @xmath158+u_{m}(t-\delta t(t))\ ] ] takes place with some coefficient @xmath159 . characteristic time scale of variations in @xmath160 , in amplitude shift @xmath161 and in the delay @xmath162 too , are about ten times slower then variations in phase and amplitude of @xmath163 during a cycle of soliton production . hence , the conclusion arises that some low - frequency component of the chaotic dynamics is not sensitive to synchronizing signal and therefore hinders to synchronize fast components which themselves are rather sensitive . additional surprising evidence for this is that sharp switching from one master signal , @xmath142 , to another ( may be the same but shifted by a time ) temporarily improves synchronization which then becomes spoiled at longer time scale . synchronization by feedback signal . seemingly , better synchronization will be achieved if synchronizing signal is taken from the `` heart '' of generator , i.e. from the feedback device . in the simplest variant shown by plot ( a ) in fig.23 , the master feedback voltage mix up with the slave one , with some coefficient @xmath164 . in this experiment besides the control voltages , @xmath143 and @xmath165 , were under comparison taken from wire inductors placed between active and passive loops ( see plot ( b ) ) . at @xmath166 , best results of synchronization with respect to the power absorption , @xmath167 , amplitude of control voltage , @xmath168 , and amplitude and instant frequency of feedback voltage , @xmath163 , are presented by four center plots in fig.23 . again , here slow time - varying delay between slave and master signals and slowly difference in their magnitudes were found , in correspondence with eq.5 . similar correlation between difference in power absorption and difference in film energies is detected in plot ( c ) . plot ( f ) allows to see what the events develop in a concrete point marked by star at plot ( b ) ( which itself represents a snapshot of quasi - local energy pattern ) . the new interesting observation follows from the histogram ( probability distribution ) of delay , at plot ( e ) . sharp peaks in this plots are separated by @xmath169 ns which is close to period of precession . this may be hint on fine features of synchronization still to be understood . clearly , a kind of rough synchronization takes place , but in details it looks even worse than in the above circumferential case . therefore stronger mixing with @xmath170 was tested . its results are shown in figs.24a - b ( see captures to these figures ) . according to top four plots in fig.24a , results occurs definitely better than at @xmath166 . this improvement is directly connected to essentially smaller difference between carrier phases of @xmath171 and @xmath172 ( plot ( 4 ) in fig.24b ) than in previous case ( plot ( d ) in fig.23 ) . .... .... 7.16 . more about defects of synchronization . from the power and feedback voltage time series it is seen that most of maxima and minima corresponding to soliton births are well reproduced by slave system . however , synchronization of events in the interspace between driving and feedback inductors is rather bad as registered by the powers , @xmath167 , and control voltages , @xmath168 , in fig.24a . this means that magnetization pattern in slave film imitates that in master film but never copies it . this misalignment can be explained as follows . there is no rigid connection between film energy and a number of solitons . some background ( smoothly distributed and slowly varying ) part of energy behaves more or less autonomously and remains out of synchronization . plots ( 1)-(5 ) in fig.24b demonstrate that all the characteristics of instant ( time - local ) quality of synchronization are closely correlated with difference between the energy backgrounds in slave and muster films . plot ( c ) shows that ( in concrete case under consideration ) characteristic frequency of chaotic variations of energy background is about 30 mhz , which is of order of inverse relaxation time , @xmath173 . fig.24c presents one more example of the same kind of synchronization , with @xmath174 , at wide film ( @xmath175 ) in strongly chaotic regime . under corrections to influence by non - synchronized energy background , the picture of synchronization with respect to power absorption looks enough good ( plot ( d ) ) . but the whole picture has no significant differences from previous case . magnetic turbulence . the figs.25a - b illustrate what happens at too powerful auto - generation which is achieved under weak output saturation , @xmath176 ma , although at relatively small linear gain ( = 4 ) . in spite of that film is narrow ( @xmath177 ) , extremely chaotic shapeless energy pattern is formed by irregularly scattered solitons ( see top picture in fig.25b ) . in this case ( which may be named turbulent magnetic chaos ) , noticeable chaotic drift ( diffusion ) of the carrier phase takes place due to phase slips , as shown at plot ( a ) in fig.25a . plots ( f)-(h ) and ( c)-(e ) present well expressed slips in local magnetization and in such the integral characteristics as feedback voltage . according to plot ( b ) , corresponding slow chaotic variations of carrier frequency are excellently anti - correlated with the energy ( as the non - isochronity implies ) two interest things should be appointed . \(i ) in contrast to commonly strong chaos , with @xmath178if estimated from feedback voltage time series ( see plot ( c ) in fig.25b ) , the energy behaves much more regularly and can be characterized by own fractal dimension , @xmath179 . possibly , this is just manifestation of above mentioned autonomy of the energy background . \(ii ) very slow chaotic variations in energy ( and thus in the carrier frequency and other related values ) are observed , with characteristic frequencies down to 3 mhz ( plots ( d ) and ( e ) in fig.25b ) . chaotic pulse auto - generation . slight decrease in linear gain ( from 4 to 3 ) at the same film at same output saturation results in new interesting regime of auto - generation which is characterized by sharp pulses of the power absorption,@xmath180 , and voltages in feedback and driving loops , @xmath181 and @xmath182 . this example is illustrated by figs.26a - b . theoretically , @xmath183 @xmath184 , where @xmath185 and @xmath186are currents in feedback and driving loops ( see scheme in fig.15 ) , and @xmath187 means averaging over period of precession . the part @xmath188 is power absorption from driving loop which can be written as @xmath189 @xmath190 , with @xmath191 and @xmath192being envelopes and @xmath193 and @xmath194 phases . fat curve and thin curve close to it at plot ( 5 ) in fig.26b show pulsations of @xmath195 and @xmath196 , respectively , while the low - amplitude curve here represents @xmath197 and shows that the feedback loop is consuming film s energy . in this case , again ( i ) essential difference between common chaos ( with @xmath198 ) @xmath8 and particular energy chaos ( with @xmath199 ) , and ( ii ) slow energy variations ( with characteristic frequencies @xmath200 mhz if not less ) @xmath8take place . at last , due to relatively simple and expressively shaped chaos and relatively small fractal dimension in this case , we may try do draw three - dimensional projection of corresponding chaotic attractor . two variants of the projection are shown in fig.28 . 7.19 . on power threshold of chaos . in numerical simulations , we can set any gain and output power as we want . in practice , small - sized high - frequency amplifiers have rather limited output power . therefore , consider rough analytical estimate of the smallest ( threshold ) power absorption by film which is sufficient for chaotic auto - generation . time averaging of the energy conservation law ( 6.3 ) , if combined with the relations ( 6.4 ) ( at @xmath201 as it was seen ) and ( 6.6 ) ( i.e. for not too short - wave chaos ) , yields @xmath202 where @xmath203is the wave function considered in sec.5 , @xmath105 is eccentricity of uniform precession , and @xmath204 is mean power absorption per unit volume ( in dimensionless units ) . hence , in according with the estimates of the minimum @xmath205 in real units , for the total threshold power absorption , @xmath206 , in a film with volume @xmath207 this formula implies @xmath208 where @xmath209is uniform precession frequency ( in @xmath8s@xmath210 ) and @xmath211remains dimensionless . in yig film , at @xmath212oe(i.e.@xmath213 in dimensionless form ) @xmath214 and @xmath215s@xmath216 . if take @xmath217 @xmath218 , which corresponds to approximately 1 oe half - width of linear fmr , and the volume be @xmath219mm@xmath8 @xmath220 mm @xmath221micron then eq.8 gives @xmath222mw . however , the estimate is rather sensitive to energy losses , and at 0.5 oe half - width of fmr it turns into @xmath223 mw . probably , this is still overestimated value since not the whole volume must be critically excited at threshold pump . we have reported selected results of work which was made in 2002 - 2003 years . additional information about this work , - in the form of html presentation ( also made at that time ) with figures , movies and comments to them , - can be obtained by downloading and unzipping file http://yuk-137.narod.ru/experience/mfs_2002.zip . this presentation includes also some information about our ( together with our colleagues ) first attempts of practical generation and synchronization of magnetic - wave chaos and its properties in comparison with that observed in numerical simulations . on the latter subjects and , besides , physical properties of our ferrite ( yig ) films and first practical applications of electro - magnetic chaos , generated with the help of these films , to direct microwave - frequency secure communication , see also our recent journal articles [ 1,2 ] . \(i ) there are two main machanisms of the magnetic - wave chaos auto - generation . one is creation of non - regular chains of ( `` gray '' ) envelope solitons composed ( in in - plane magnetized films under our above consideration ) primarily with ` surface' ( `` damon - eshbach '' ) magnetostatic spin waves ( msw ) . another mechanism is parametric energy transfer from these waves to relatively short `` backward '' msw ( `` bulk '' msw ) and reversed process . \(ii ) the first mechanism dominates at not too large ( in - plane ) magnetizing field ( therefore , generates lower carrier frequencies ) but it needs in much smaller level of energy consumption ( dissipation ) in film and , hence , much smaller amplification in a feedback circuit . besides , importantly , it produces much more `` rich '' chaos , which allows , in principle , to `` hide '' a large amount of information . \(iii ) however , these advantages of the `` parametric '' chaos ( as compared with the `` solitonic '' one ) are accompanied with probable instability of the parametric interaction between surface and backward msw , which may take form of explosive and irreversible transfer of energy from surface to short backward waves . a degree of the irreversibility is as large as great is number of excited backward wave modes per one exciting surface mode . at some regimes of auto - generation one can observe from time to time fast and practically full energy swaps into short backward waves and , consequently , sharp and long - standing suppressions of auto - generation ( sometimes , under definite conditions , it may break for periods up to nearly @xmath224s ) . this phenomenon ( in the presentation , we called it `` parametric collapse '' ) was observed both in numerical and real experiments ( with good correspondence between them , though numeric procedures inevitably loss most short waves ) . naturally , it can create strong obstacles to synchronization of chaos , because the breaks of auto - generation imply uncontrollable and hardly reproducible in `` slave '' system ) phase slips . ( iv ) therefore , to avoid the `` parametric collapse '' but , at the same time , exploit parametric processes as much as possible , it is necessary to introduce some auto - regulation of feedback amplification . it can help in generation of such output electromagnetic signals what possess not too strong amplitude modulation , at strong enough phase and frequency chaotic modulation , and cause satisfactory `` synchronous chaotic response '' in slave system . \(v ) fractal ( correlation ) dimensions of both high - frequency chaotic oscillations of different variables ( local magnetization values , e.m.f.s in antennae , and voltages and currents in feedback circuit ) and low - frequency variables ( amplitudes , phases and `` instant frequencies '' ) almost always lie in interval from 2 to 4 and usually between 3 and 4 , thus demonstrating `` hyper - chaos '' . this numeric result well agrees with fractal dimensions of low - frequency ( amplitude and phase ) signals obtained from real chaotic generators . at the same time , some of low - frequency variables , both real and numerical ( e.g ` quasi - local energy density' and total `` excess '' magnetic energy ) can show dimensions nearly by unit lesser , from interval 2@xmath2253 . ( vi ) these facts , seemingly , help to understand why , in spite of not high fractal dimension of the our chaos , we never observe `` ideal synchronization '' , instead seeing only what can be termed `` generalized synchronization '' . the matter is that , in the schemes under investigation , all the control and feedback signal are directly coupled with the `` surface '' ( `` damon - eshbach '' ) msw modes only . this is obvious from comparison of voltage - current spectra , - whose frequency band always lies near or higher the uniform precession frequency , - and numerically found spectra of local magnetization values ( local precession ) which can be concentrated mainly at twice lower frequencies or even four times lower ones ( because of two successive parametric divisions of frequency ) , down to absolute lower bound of msw spectrum ( moreover , sometimes , - e.g. after the parametric collapse , - become concentrated just near this bound ) . therefore , it is not surprising if one of 4 effective relevant degrees of freedom of our system ( the number 4 follows from fractal dimension between 3 and 4 ) , - which is responsible for magnitude of short backward msw and their energy exchange with surface msw , - stays far out of control we used . if this is true explanation of the generalized character of synchronization , then we have chances to improve the described results .	selected results of original numeric simulations of non - linear magnetostatic spin waves and microwave - frequency magnetic chaos in ferrite films are expounded , as third part of the work whose first two parts are recent arxive preprints 1204.0200 and 1204.2423 . especially we consider crucial role of parametric processes in creating the chaos and simultaneously obstacles to its synchronization , and examine some possibilities of good enough synchronization ( to an extent allowing its use for direct secure communication in microwave band ) .
the source grs 1758258 was discovered in the hard x ray / soft @xmath3ray energy range with the sigma / granat coded mask telescope ( sunyaev et al . grs 1758258 is of particular interest since , together with the more famous source 1e 1740.7 2942 , it is the only persistent hard x ray emitter ( @xmath0 kev ) in the vicinity of the galactic center ( goldwurm et al . both sources have peculiar radio counterparts with relativistic jets ( mirabel et al . 1992a ; rodriguez , mirabel & mart 1992 ; mirabel 1994 ) and might be related to the 511 kev line observed from the galactic center direction ( bouchet et al . 1991 ) . despite the precise localization obtained at radio wavelengths , an optical counterpart of grs 1758258 has not been identified ( mereghetti et al . 1992 ; mereghetti , belloni & goldwurm 1994a ) . simultaneous rosat and sigma observations , obtained in the spring of 1993 , indicated the presence of a soft excess ( mereghetti , belloni & goldwurm 1994b ) . this spectral component ( @xmath4 kev ) was weaker in 1990 , when the hard x ray flux ( @xmath5 kev ) was in its highest observed state . on the basis of its hard x ray spectrum , grs 1758258 is generally considered a black hole candidate ( tanaka & lewin 1995 ; stella et al . the possible evidence for a soft spectral component anticorrelated with the intensity of the hard ( @xmath6 kev ) emission supports this interpretation . no detailed studies of grs 1758258 in the `` classical '' x ray range have been performed so far . here we report the first observations of this source obtained with an imaging instrument in the @xmath7 kev energy range . the observation of grs 1758258 took place between 1995 march 29 22:39 ut and march 30 15:38 ut . the asca satellite ( tanaka , inoue & holt 1994 ) provides simultaneuos data in four coaligned telescopes , equipped with two solid state detectors ( sis0 and sis1 ) and two gas scintillation proportional counters ( gis2 and gis3 ) . we applied stringent screening criteria to reject periods of high background , and eliminated all the time intervals with the bright earth within 40 degrees of the pointing direction for the sis data ( 10 degrees for the gis ) , resulting in the net exposure times given in table 1 . & exposure time ( s)&count rate ( counts / s ) + sis0&9,471&[email protected] + sis1&9,455&[email protected] + gis2&12,717&[email protected] + gis3&11,949&[email protected] + figure 1 shows the image obtained with the gis2 detector . most of the detector area is covered by stray light due to the bright source gx51 , located outside the field of view , at an off axis angle of about 40 arcmin . fortunately , grs 1758258 lies in a relatively unaffected region of the detector , which allows us to estimate the contamination from gx51 as explained below . the source counts were extracted from a circle of 6 arcmin radius centered at the position of grs 1758258 , and rebinned in order to have a minimum of 25 counts in each energy channel . due to the present uncertainties in the asca response at low energies , we only considered photons in the 0.810 kev range . the background spectrum was extracted from the corresponding regions of observations of empty fields provided by the asca guest observer facility . the contribution to the background due to gx51 is mostly concentrated in a circular segment with area @xmath936 arcmin2 indicated with a in figure 1 . its spectrum was estimated by the difference of regions a and b , and added to the background . a similar procedure was followed to extract the gis3 net spectrum . using xspec ( version 9.0 ) we explored several spectral models by simultaneously fitting the data sets of both gis instruments . the best fit was obtained with a power law with photon index @xmath10 and column density @xmath11 cm@xmath2 ( reduced @xmath13 for 372 d.o.f . , errors at 90% confidence intervals for a single interesting parameter ) . other models based on a single spectral component ( e.g. blackbody , thermal bremsstrahlung ) gave unacceptable results , with the exception of the comptonized disk model of sunyaev & titarchuk ( 1980 ) . however , the limited energy range of the asca data alone , does not allow in the latter case to pose interesting constraints on the fit parameters . the gis instruments have a time resolution of 0.5 s or 62.5 ms , according to the available telemetry rate . most of our data had the higher time resolution . using a fourier transform technique , and after correction of the times of arrivals to the solar system barycenter , we performed a search for periodicities . no coherent pulsations in the 0.1251000 s period range were found . for the hypothesis of a sinusoidal modulation we can set an upper limit of @xmath95% to the pulsed fraction . both sis instruments were operated in the single chip mode , which gives a time resolution of 4 s and images of a square 11x11 arcmin2 region ( figure 2 ) . most of the sis data ( 83% ) were acquired in `` faint mode '' and then converted to `` bright '' . this allows to minimize the errors due to the echo effects in the analog electronics and to the uncertainties in the dark frame value ( otani & dotani , 1994 ) . the inclusion of the data directly acquired in bright mode resulted in spectra of lower quality ( significant residuals in the 23 kev region ) . we therefore concentrated the spectral analysis on the faint mode data . the source counts ( 0.610 kev ) were extracted from circles with a radius of 3 arcmin , and the resulting energy spectra ( 1024 pi energy channels ) rebinned in order to have at least 25 counts in each bin . we subtracted a background spectrum derived during our observation from an apparently source free region of the ccds ( see figure 2 ) . this background is higher than that obtained from the standard observations of empty fields , probably due to the contamination from gx51 . it contributes @xmath94% of the extracted counts . we verified that the derived spectral parameters do not change significantly if we use the blank sky background file , or even if we completely neglect the background subtraction . by fitting together the data from the two sis we obtained results similar to those derived with the gis instruments . in particular , a power law spectrum gives photon index @xmath14 and @xmath15 cm@xmath2 , with a reduced @xmath16 of 1.031 ( 872 d.o.f . ) . no prominent emission lines are visible in the spectrum of grs 1758258 ( as already mentioned , some features in the region around 2 kev are probably due to instrumental problems , they appear stronger when the bright mode data and the corresponding response matrix are used ) . upper limits on the possible presence of an iron line were computed by adding a gaussian line centered at 6.4 kev to the best fit power law model and varying its parameters ( intensity and width ) until an unacceptable increase in the c2 was obtained . the 95% upper limit on the equivalent width is @xmath950 ev for a line width of @xmath17 kev and increases for wider lines ( up to @xmath9110 ev for @xmath18 kev ) . also in the case of the sis , a search for periodicities ( limited to period greater than 8 s ) resulted only in upper limits similar to the gis ones . the soft x ray flux observed with rosat in 1993 ( mereghetti et al . 1994b ) was higher than that expected from the extrapolation of the quasi simultaneous sigma measurement ( @xmath5 kev ) , indicating the presence of a soft spectral component with power law photon index @xmath93 below @xmath92 kev . clearly , such a steep , low energy component is not visible in the present asca data , which are well described by a single flat power law . the corresponding flux of @xmath19 ergs cm@xmath2 s@xmath20 ( in the 110 kev band , corrected for the absorption ) is within the range of values measured in march april 1990 ( sunyaev et al . 1991 ) , when the source was in its highest observed state . this fact is consistent with the presence of a prominent soft component only when the hard x ray flux is at a lower intensity level . though a single power law provides an acceptable fit to the asca data , we also explored spectral models consisting of two different components : a soft thermal emission plus a hard tail . for instance , with a blackbody plus power law , we obtained a good fit to both the sis and gis data with @xmath21 kev and photon index @xmath22 ( @xmath23 ) . obviously such a power law must steepen at higher energy to be consistent with the sigma observations . in fact a sunyaev titarchuk comptonization model can equally well fit the asca hard tail and provide an adequate spectral steepening to match the high energy data ( see figure 3 ) . good results were also obtained when the soft thermal component was fitted with models of emission from accretion disks ( e.g. makishima et al . 1986 , stella & rosner 1984 ) . in all cases the total flux in the soft component amounts only to a few percent of the overall ( 0.1 300 kev ) luminosity . however , the low observed flux , coupled to the high accretion rates required by the fitted temperatures , implies an unplausible large distance for grs 1758258 and/or very high inclination angles ( note that there is no evidence so far of eclipses or periodic absorption dips which could hint to a high inclination system ) . a possible alternative solution is to invoke a significant dilution of the optically thick soft component by comptonization in a hot corona . a very rough estimate shows that , in order to effectively remove photons from the thermal distribution , a scattering opacity of @xmath24 is required . our asca observation provides the most accurate measurement of the absorption toward grs 1758258 obtained so far . obviously the derived value is slightly dependent on the adopted spectral model . however , values within at most 10% of @xmath25 cm@xmath2 were obtained for all the models ( one or two components ) fitting the data . this column density is consistent with a distance of the order of the galactic center and similar to that of other sources in the galactic bulge ( kawai et al . 1988 ) , but definitely smaller than that observed with asca in 1e 1740.72942 ( churazov et al . 1996 ) . the information on the galactic column density , coupled to the optical / ir data , can yield some constraints on the possible companion star of grs 1758258 ( see chen , gehrels & leventhal 1994 ) . a candidate counterpart with @xmath2619 and @xmath2717 ( mereghetti et al . 1994a ) lies within @xmath92 `` of the best radio position ( mirabel et al . other infrared sources present in the x ray error circle ( 10 '' radius ) are fainter than @xmath28 ( mirabel & duc 1992 ) . using an average relation between @xmath29 and optical reddening ( gorenstein 1975 ) , we estimate a value of @xmath30 , corresponding to less than one magnitude of absorption in the k band ( cardelli , clayton & mathis 1989 ) . thus , for a distance of the order of 10 kpc , the k band absolute magnitude must be fainter than @xmath31 . this limit clearly rules out supergiant or giant companion stars , as well as main sequence stars earlier than type a ( johnson 1966 ) , thus excluding the possibility that grs 1758258 is in a high mass binary system . the flux of grs 1758258 measured with the sis instruments corresponds to a 110 kev luminosity of @xmath32 ergs s@xmath20 ( for a distance of 10 kpc ) . a reanalysis of archival data from ttm / mir , xrt / spacelab2 and exosat ( skinner 1991 ) , showed that grs 1758258 had a similar intensity also in 1985 and in 1989 . an earlier discovery had been prevented only by confusion problems with gx51 , much brighter than grs 1758 258 below @xmath920 kev . subsequent hard x ray observations with sigma ( gilfanov et al . 1993 , goldwurm et al . 1994 ) repeatedly detected grs 1758258 with a hard spectrum extending up to @xmath9300 kev . it is therefore clear that grs 1758258 , though variable by a factor of @xmath910 on a timescale of months , is not a transient source . the asca satellite has provided the first detailed data on grs 1758258 in the 110 kev region , allowing to minimize the confusion problems caused by the vicinity of gx51 , that affected previous observations with non imaging instruments . the possible black hole nature of grs 1758258 , inferred from the high energy data ( sunyaev et al . 1991 , goldwurm et al . 1994 ) , is supported by the asca results . the power law spectrum , extending up to the hard x ray domain is similar to that of cyg x1 and other black hole candidates in their low ( or hard ) state . furthermore , our stringent limits on the presence of periodic pulsations and accurate measurement of interstellar absorption make the possibility of a neutron star accreting from a massive companion very unlikely . the lack of iron emission lines in the sis data has to be confirmed by more stringent upper limits to rule out , e.g. , the presence of a reflection component as proposed for cyg x1 ( done et al . 1992 ) . for comparison , the iron line recently observed with asca in cyg x 1 has an equivalent width of only 1030 ev ( ebisawa et al . 1996 ) . the prominent soft excess observed with rosat in 1993 , when the hard x ray flux was in a lower intensity state , was absent during our observation . the source was in a hard spectral state , with a possible soft component accounting for @xmath95% of the total luminosity at most . a similar soft component ( @xmath33 kev ) , but contributing a larger fraction of the flux , has been observed in cyg x1 and attributed to emission from the accretion disk ( balucinska church et al . if the soft component in grs 1758258 originates from the disk , strong dilution is required . an optically thick hot cloud embedding the innermost part of the disk is an attractive hypothesis . to test the viability of this model , a detailed fit to simultaneous data over a broad energy range , as available , e.g. , with sax in the near future , is required .	grs 1758258 is one of the few persistent hard x ray emitters ( @xmath0 kev ) in the galaxy . using the asca satellite , we have obtained the first detailed data on grs 1758258 in the 110 kev range , where previous observations were affected by confusion problems caused by the nearby strong source gx51 . the spectrum is well described by a power law with photon index 1.7 without strong fe emission lines . a prominent soft excess , as observed with rosat when the hard x ray flux was in a lower intensity state , was not detected . however , the presence of a soft spectral component , accounting for at most 5% of the 0.1300 kev flux , can not be excluded . the accurate measurement of interstellar absorption ( @xmath1 cm @xmath2 ) corresponds to an optical extinction which definitely excludes the presence of a massive companion .
a long term radio monitoring study of the @xmath4ms radio pulsar psrb1931 + 24 ( stokes et al . 1985 ; hobbs et al . 2004 ) , revealed the peculiar intermittent behavior of this pulsar ( kramer et al . psrb1931 + 24 is ( so far ) a unique system : it shows an active radio emission phase lasting between 510 days ( radio on phase , hereafter ) , which suddenly ( in less than 10s ) switches off , and the pulsar remains undetectable for the following 2535 days ( radio off phase , hereafter ) . this pattern repeats quasi - periodically , and has been monitored for the past 7 years ( kramer et al . another peculiar property of psrb1931 + 24 is its spin down behavior . remarkably , the pulsar rotation slows down considerably faster ( by about 50% ) when the pulsar is in the radio on phase , with a frequency derivative changing from @xmath5s@xmath6 to @xmath7s@xmath6 across the two phases . from the radio observations , typical pulsar characteristics were derived ( see tab.1 in kramer et al . 2006 ) : an estimate of the dipolar magnetic field ( @xmath8 g ) , the characteristic age ( @xmath9myr ) , and the dispersion measure ( @xmath10@xmath11pc ) . the latter gave a rough estimate of the pulsar distance of @xmath12kpc using the ne2001 model for the free electron distribution in the galaxy ( cordes & lazio 2002 ) . kramer et al . ( 2006 ) interpreted the peculiar spin - down properties of this pulsar in terms of a transient plasma flow in the magnetosphere , causing the quenching and re - ignition of the radio emission . in particular , an increase of the magnetospheric current flow during the radio - on phase , can provide an additional braking torque on the neutron star . from the observed variations in the pulsar spin - down , these authors showed that the pulsar magnetospheric electron density was consistent with the goldreich - julian density ( goldreich & julian 1969 ) . this interpretation can successfully explain the variable neutron star torque , but still fails to explain what causes the changes in the magnetospheric plasma flow , especially if in a quasi periodic fashion . to date , this periodicity of the radio on and radio off recurrence is difficult to explain in any scenario considering an isolated pulsar . as kramer et al . ( 2006 ) pointed out , the short shut - off time of less than 10 seconds , and the rather stable pulse profile , rule out possible scenarios like precession of the neutron star . on the other hand , the long radio off phase is in contrast with the typical nulling timescales of radio pulsars , exceeding it by almost five orders of magnitude . cordes & shannon ( 2006 ) studied the possibility that psrb1931 + 24 might be in a binary system with a large asteroid , surrounded by a disk of small asteroids , moving in a @xmath13day eccentric orbit around the pulsar . this scenario suggests that the interaction between the pulsar magnetosphere and the asteroids can be responsible both for the torque change and the intermittent radio activity of psrb1931 + 24 . recently , zhang , gil & dyks ( 2007 ) have proposed that intermittent pulsars are old isolated neutron stars which entered ( or are about to enter ) the so called `` death valley '' ( chen & ruderman 1993 ) , where the polar cap voltage of the pulsar is not sufficient to power the pair production process . however , this scenario still fails to account for the quasi - periodic pulsed emission of psrb1931 + 24 . here we report on x ray ( [ chandra ] ) and optical ( [ optical ] ) observations of psrb1931 + 24 taken with the _ chandra x ray observatory _ and the _ isaac newton telescope_. furthermore , in [ orbital ] we derive constraints on possible orbital parameters and companion mass by studying the radio timing residuals . in [ binary ] and [ discussion ] , we discuss our results and investigate in detail the possibility that psrb1931 + 24 resides in a binary system with a @xmath035day orbital period and a low mass companion star or a planet . the _ chandra _ advanced ccd imaging spectrometer ( acis ) observed psrb1931 + 24on 2006 july 20th , for an on source exposure time of @xmath09.7ks . the target position was placed in the standard back illuminated acis s3 aimpoint , using the faint mode . we corrected the astrometry for any processing offset and we cleaned the image for hot pixels . running ciao celldetect and wavedetect tools , no x - ray sources were detected in the whole acis s3 ccd , while four ( unrelated ) sources were detected in the other ccds at @xmath14 confidence level over the background . in particular , no photons were detected in a @xmath15 radius around the psrb1931 + 24 radio position ( see tab.1 for the ra and dec ) . we then obtained a 99% upper limit on the source count rate of @xmath16counts@xmath2 ( gehrels 1986 ) . assuming a conservative absorption value of @xmath17cm@xmath6 ( derived for the pulsar position from dickey & lockman 1990 , hence considering the whole galactic @xmath19 in that direction ) , and using the pimms calculator , we derived a 99% upper limit on the 0.310kev unabsorbed flux of @xmath20ergs@xmath2cm@xmath6 or @xmath21ergs@xmath2cm@xmath6 , assuming a black body ( kt = 0.3kev ) or a power law ( @xmath22 ) spectral decomposition , respectively . at 4.6kpc these fluxes translate into a 99% upper limit on the 0.310kev x - ray luminosity of @xmath23ergs@xmath2 and @xmath24ergs@xmath2 , depending on the assumed x ray spectrum . however , note that the distance of 4.6kpc , as inferred from the pulsar dm ( kramer et al . 2006 ) might well have an uncertainty of a factor of 2 due to a @xmath030% uncertainty in the galactic electron density model ( cordes & lazio 2002 ) . hence , more conservative upper limits on the luminosity would be 4 times larger than reported above . we observed the field of psrb1931 + 24 with the _ isaac newton telescope _ located at the roque de los muchachos observatory in la palma , for an exposure time of 10 minutes in three optical filters : g@xmath25 , r@xmath25 and i@xmath25 ( see fig . [ opt_obs ] for the finding - chart ) . no optical counterpart was detected within the radio pulsar position . we derived 5@xmath26 upper limits on the optical emission of psrb1931 + 24of g@xmath27 , r@xmath28 and i@xmath29 magnitudes . we inferred these optical upper limits from the magnitudes of the faintest object detected at 5@xmath26 confidence level in the same ccd as the pulsar position . for the astrometry we used an @xmath30 subsection of the g@xmath25 image where we found 85 stars from the usno ccd astrograph calalogue ( ucac2 ; zacharias et al . we obtained an astrometric solution , fitting the zero point position , scale and position angle ; the final rms residuals were @xmath31 in both in ra and dec . .parameters derived from the radio timing analysis described in [ orbital ] . errors are at 1@xmath26 confidence level . [ cols="<,>",options="header " , ] we inferred the reddening in the direction of psrb1931 + 24 from the @xmath19 , which gave an @xmath32 ( predehl & schmitt 1995 ) . we then converted this value into an estimate of the reddening in the filter we actually used ( rieke & lebofsky 1985 ; schlegel et al . 1998 ) : @xmath33 , @xmath34 and @xmath35 . considering a distance of 4.6kpc , we inferred upper limits on the source absolute magnitude of : g@xmath36 , r@xmath37 and i@xmath38 . from our upper limits of the absolute optical magnitudes , we could derive a rough range of stellar types for a putative counterpart . from r@xmath39 , which was the most constraining limit , we could exclude all giant and super - giant stars , and deriving a spectral type later than a g8 star . note that having we assumed the absorption value of the whole galactic @xmath19 : any ( more realistic ) lower @xmath19 values would result in stellar types much later than a g8 . if psrb1931 + 24 happens to be in a binary system ( e.g. as investigated in detail in [ binary ] ) , the motion of the pulsar around the system s centre of mass should leave a periodic signature in the remaining timing residuals . we searched for this signature in radio timing data spanning almost 7 years ( kramer et al . we fit the radio timing residuals with one spin frequency derivative and a second derivative ( values are reported in tab.1 ) . the latter was used to remove a cubic term due to long term timing noise . in order to improve the position information , we removed the remaining timing noise by using the fit - wave method described by hobbs et al . ( 2004 ) , with a minimum period of 2 years . the pre - fit - wave residuals were then analysed for periodicities caused by low - mass companions ( with masses smaller then the pulsar mass , @xmath40 ) , as described by freire et al . ( 2003 ) . as fig.2 shows , we detected , unsurprisingly , a faint signal at a period of @xmath41 days . however , we believe this signal is consistent with a periodicity caused by the sampling of time - of - arrivals forced upon us by the quasi - periodic on - phases of the pulsar . we do not detect any other significant signal . for a more general companion mass we also performed the following analysis of the timing residuals . the peak - to - peak amplitude of the remaining timing residuals was @xmath422ms . from this we placed limits on the corresponding orbital parameters by interpreting @xmath43 as caused by a `` roemer delay '' , i.e. the light - travel time across the orbit ( e.g. lorimer & kramer 2005 ) . we then set : @xmath44\ ] ] equal to @xmath43 . here , @xmath45 is the eccentric anomaly , @xmath46 the eccentricity and @xmath47 the longitude of the periastron . furthermore , @xmath48 is the projected semi - major axis measured in light - seconds , which is a function of @xmath49 , the semi - major axis of the orbit is the orbital inclination angle , @xmath50 is the speed of light , and @xmath51 is the size of the relative orbit . ] . on the other hand , the relative orbit @xmath51 , the orbital period and masses are related according to kepler s 3rd law : @xmath52 with the masses measured in solar units and @xmath53s . assuming @xmath54days and @xmath55 we could derive an estimate for the orbital inclination angle as a function of the companion mass in the following way : for a given companion mass @xmath56 we performed monte - carlo simulations drawing possible values for @xmath45 , @xmath47 and @xmath46 from uniform distributions over @xmath57 $ ] and @xmath58 $ ] , respectively . note that the distribution for the eccentricity is not expected to be uniform ( i.e. it is likely to be skewed to small eccentricities ) but for our purposes this assumption is sufficient as we were mostly interested in upper bounds for the inclination angle . for each companion mass , one million monte - carlo runs were performed and the median and 95% confidence limits were computed ( see fig . [ radio ] : the grey shadowed region is the allowed region at 95% confidence level ) . hence , for reasonable companion masses the tight limit on a detectable periodicity in the timing residuals implies rather small orbital inclination angles ( see also [ discussion ] ) . in this section we investigate the possibility that the peculiar characteristics of psrb1931 + 24 can be explained by a low mass star or a gaseous planet moving in an eccentric orbit around the pulsar ( see also [ orbital ] ) . this causes a quasi - periodic intermittent behavior , as well as the change in the spin - down characteristics of the pulsar , in two different ways : 1 ) providing at the periastron passage the additional plasma needed in the kramer et al . ( 2006 ) model , or 2 ) accreting material on the pulsar magnetosphere and causing the transition between the radio pulsar and the propeller regime ( see for details illarionov & sunyaev 1975 ; stella et al . 1986 , 1994 ) . in this section we discuss only this second possibility . the physics involved in the first possibility has been thoroughly investigated by kramer et al . ( 2006 ) , although without considering the wind of a companion as the source of the additional magnetospheric plasma during the radio on phase ( see also [ discussion ] ) . we define here three important psrb1931 + 24 radii which will be used in the following . the magnetospheric radius : @xmath59 the corotation radius : @xmath60 and the light cylinder radius : @xmath61 these three radii represent , respectively : i ) the balance of the gravitational force and the ram pressure of the infalling material , ii ) the centrifugal barrier for the infalling material due to the pulsar rotation , and iii ) the place where the pulsar magnetic field lines open such that their tangential velocity do not exceed the speed of light ( illarionov & sunyaev 1975 ; ruderman & sutherland 1975 ) . we define : @xmath62 g is the neutron star magnetic field , @xmath63 is the spin period in units of 10ms , @xmath64 g s@xmath2 is the mass inflow rate toward the neutron star , and @xmath65 and @xmath66 cm are the mass and the radius of the neutron star , hereafter assumed @xmath67 . note that during the orbital motion , the only variable radius is @xmath68 . if the mass inflow rate from the putative companion star is very low , as might happen at apastron of an eccentric orbit , the magnetosphere of the neutron star might be larger than the light cylinder and corotation radii @xmath69 . as a result , the centrifugal barrier is closed , which means that incoming material from the companion star is prevented from falling towards the neutron star magnetosphere or surface . in this case the neutron star might behave as a radio pulsar . the pulsar radiation pressure dominates over the ram pressure of the inflowing material , preventing the matter to penetrate towards the neutron star ( illarionov & sunyaev 1975 ; davis & pringle 1981 ; stella et al . 1994 ; campana et al . 1998 ) . this occurs when the mass inflow rate ( @xmath70 ) is smaller than the limiting value : @xmath71 above which the radio pulsations would quench , even as rapidly as @xmath010s . it is worth noting that when @xmath72 , the radio pulsar emission is not in equilibrium ( illarionov & sunyaev 1975 ; shaham & tavani 1991 ) , depending on the stability of the mass inflow rate ; sporadic variability of the radio on duration of psrb1931 + 24 might take place , as indeed observed ( kramer et al . furthermore , during the radio on phase , variations in the pulsar dm as a function of the orbital phase are expected . unfortunately the observations were performed at one frequency , making impossible to put a meaningful limit on the dm variability ( see also kramer et al . 2006 ) . if the mass inflow rate starts to increase , e.g. approaching periastron , we expect a correlated decrease of the magnetospheric radius , which eventually becomes smaller than the light cylinder radius @xmath73 . this happens when the mass inflow rate towards the neutron star becomes larger than the limiting value @xmath74 ; at this point the pulsar radiation pressure is overcome by the ram pressure of the infalling material quenching the radio pulsar mechanism . different regimes are then allowed at this stage . if @xmath75 remains larger than @xmath76 , which means @xmath77 , with @xmath78 then the magnetosphere of the neutron star still rotates in a super - keplerian motion , and the inflowing material might either accumulate outside the magnetospheric boundary or be swept away by the magnetospheric drag : this is called the `` propeller '' regime ( pringle & rees , 1972 ; illaroniov & suniaev 1975 ; davies & pringle 1981 ; wang & robertson 1983 ; stella , white & rosner 1986 ) . given the upper limits we derived for the possible companion star ( see [ optical ] and [ orbital ] ) , there is only a very small chance that the @xmath70 , due to the wind loss of the companion , overcomes @xmath79 . we will then consider hereafter only the possibility that the radio off phase is driven by the propeller regime , excluding the surface accretion scenario . when the radio pulsations are quenched , the spin down behavior of the pulsar is not driven anymore by the magnetic dipolar loss as it was before . what happens to the pulsar spin down during the propeller regime is still rather controversial , and requires detailed hydrodynamic simulations ( romanova et al . depending on which kind of instability and shock takes place on the pulsar magnetosphere , the spin down rate might either increase further or be reduced by the angular momentum transferred by the infalling material to the magnetosphere . in the psrb1931 + 24 case , it is clear that the infalling material should provide a certain rotational energy in order to make the pulsar slow down less efficiently during the radio off phase , changing the spin - down of the neutron star by @xmath80 hzs@xmath2 , which converted in energy corresponds to @xmath81 . when the infalling matter reaches the neutron star magnetosphere , the source is expected to emit in the x ray band with a minimum luminosity of : @xmath82 and a maximum luminosity related to the maximum wind loss we expect from the inferred companion type . for a stellar type later than a g8 , we expect the x ray luminosity of the system in the radio off phase to be between @xmath83ergs@xmath2 . after the periastron passage , the low - mass companion star moves towards apastron , the magnetospheric boundary expands because of the decreasing mass inflow rate , the centrifugal barrier closes and the radio pulsar mechanism can resume again . the upper limits we derived from these x - ray and optical observations of psrb1931 + 24 during the radio on phase , are not deep enough to detect the typical x ray luminosity ( @xmath84erg s@xmath2 ; becker & trmper 1997 ) and optical magnitude ( v@xmath8528 ; mignani et al . 2007 ) of an isolated pulsar of @xmath01.6myr , as psrb1931 + 24 . on the other hand , if the pulsar is hosted in a binary system , we derived tight constraints on the putative companion and its orbital parameters . we constrained the companion to be a low mass star later than a g8 type or a gaseous planet . for an orbital inclination of 0.1 degree the companion would have a mass @xmath860.01 m@xmath87 , while with a smaller inclination of 0.01 degree , the companion mass is allowed to be @xmath860.1 m@xmath87 ( all at 95% confidence level ; see fig.3 ) . in the binary scenario ( see [ binary ] ) , the pulsar intermittence can be explained by the transition between the radio pulsar regime and the propeller regime during the orbital motion . the x ray upper limits we derived are unfortunately relative to the radio pulsar dominated regime ( note that the source was radio on during the observations ) , and not to the propeller regime when the source can be bright in the x - ray due to accretion of matter onto the pulsar magnetosphere . our _ chandra _ observation was in fact aimed at detecting a possible x ray emission from psrb1931 + 24 during its radio off phase . unfortunately , the relatively uncertain duration of both phases ( the radio on phase rarely might happen to be longer than the average duration ) , and the relatively large error ( @xmath0 a few days ) on the putative periastron passage , made the observation be performed during the radio on phase of the pulsar . note that the insufficient flux detection limit of current and past x ray surveys and monitoring programs ( e.g. rosat and the rxte / asm ) , do not make the non detection of an x ray counterpart in previous x - ray surveys at all constraining . although very successful in explaining the intermittence , the quasi periodicity and the torque variations of psrb1931 + 24 , this binary scenario still presents several unsolved issues . one issue of this scenario , and to some extent also to the `` asteroid '' model ( cordes & shannon 2006 ) , is the very small orbital inclinations required by the radio timing studies ( [ orbital ] and fig.3 ) . for a random distribution of inclination angles , the probability of observing a system at an angle less than a value @xmath88 , is p@xmath89 . considering the favorable case of an angle of 15@xmath90 ( a gaseous planet ; 3 ) , the probability of observing such a system is p@xmath91% . a chance of 3.4% is not very high , although still worth to be considered . another issue concerns the accretion rate from the companion . can the accretion from the low wind loss of such a small mass star or gaseous planet be high enough to switch the pulsar from the radio emitting regime to the propeller regime ? typical wind rates for e.g. a k0 star are insufficient to produce the limiting value of @xmath74 by @xmath02 orders of magnitude . this problem might be partially ( but not totally ) alleviated by taking into account the irradiation process ( podsiadlowski 1991 ; dantona & ergma 1993 ) . in particular , a low mass star irradiated by the radio pulsar is expected to expand even if the radiation bath is not particularly extreme , and the wind of the star to increase substantially . however , the irradiation process can not be dominant on the stellar wind from a g8 to k0 star , mainly because of the relatively low rotational energy of psrb1931 + 24 with respect to the surface temperature of the companion star ( @xmath92 , where @xmath93 is the companion star radius ) . for lower mass stars or considering a gaseous planet orbiting around the pulsar , this process might have instead a substantial role , and might produce strong winds towards the pulsar while at periastron . if the weak stellar wind from the low - mass companion star is insufficient to drive the radio pulsar in the propeller regime , it is instead strong enough to provide the amount of plasma needed to produce the spin down change in the scenario proposed by kramer et al . note that in this picture , at the periastron passage we expect the radio on phase , while in the propeller scenario it is expected at the apastron of the orbit . however , while the propeller scenario naturally explains the quenching of the radio emission , in this picture the radio quenching still remains puzzling : we would in fact expect to see the pulsar at all the time , although with two different spin down behavior . we acknowledge the director of the chandra x - ray observatory , harvey tananbaum , for promptly according us an observation of psrb1931 + 24 through his director discretionary time , and for useful comments on this draft . we also thank the chandra team for the efficiency during the scheduling process . we are grateful to the referee , s. ransom , for his valuable suggestions which largely improved our work . nr acknowledges support from an nwo veni fellowship and thanks f. verbunt , m. burgay , d. russell , and the atnf epping pulsar group , for useful comments and discussions .	psrb1931 + 24 is the first intermittent radio pulsar discovered to date , characterized by a 0.8s pulsation which turns on and off quasi - periodically every @xmath035days , with a duty cycle of @xmath010% . we present here x ray and optical observations of psrb1931 + 24 performed with the _ chandra x - ray observatory _ and _ isaac newton telescope _ , respectively . simultaneous monitoring from the _ jodrell bank observatory _ showed that this intermittent pulsar was in the radio on phase during our observations . we do not find any x ray or optical counterpart to psrb1931 + 24 , translating into an upper limit of @xmath1ergs@xmath2 on the x ray luminosity , and of @xmath3 on the optical magnitude . if the pulsar is isolated , these limits can not constrain the dim x - ray and optical emission expected for a pulsar of that age ( @xmath01.6myr ) . we discuss the possibility that the quasi periodic intermittent behavior of psrb1931 + 24 is due to the presence of a low mass companion star or gaseous planet , tight with the pulsar in an eccentric orbit . in order to constrain the parameters of this putative binary system we re - analysed the pulsar radio timing residuals and we found that ( if indeed hosted in a binary system ) , psrb1931 + 24 should have a very low mass companion and an orbit of low inclination . [ firstpage ] stars : pulsars : general pulsar : individual : psrb1931 + 24
based on several astronomical evidences , our universe is currently in a period of positively accelerated expansion [ 1 ] . it is possible to interpret this late - time acceleration based on yet unknown component called dark energy in literature ( see for instance [ 2 ] with a comprehensive list of references therein ) . also , it has been shown that such an accelerated expansion could be the result of a modification to the einstein - hilbert action ( for a recent review see [ 3 ] ) . on the other hand , dgp braneworld scenario has the capability to interpret this late - time acceleration via leakage of gravity to extra dimension in its self - accelerating branch [ 4 ] . for the first alternative , the simplest candidate for dark energy is the cosmological constant itself . however , it suffers from serious problems such as a huge amount of fine - tuning [ 2,5,6 ] . within dark energy viewpoint , for a dark energy component , say @xmath1 , its equation of state or equivalently equation of state parameter @xmath2 , determines both gravitational properties and evolution of the dark energy . recent constraints on equation of state of dark energy indicate that @xmath3 and even that @xmath4 . when we consider background evolution , there is no problem with @xmath4 . however , there will be apparent divergencies in perturbation when one crosses the phantom divide line , @xmath5 . accordingly , crossing of the phantom divide line , @xmath5 , provides a suitable basis to test alternative theories of gravity or higher dimensional models which can give rise to an effective phantom energy [ 6 ] . with these motivations , the issue of phantom divide line crossing has been investigated extensively in recent years ( see [ 6 ] and references therein ) . in principle , crossing of phantom divide line by the dark energy equation of state parameter at recent red - shifts , has two possible cosmological implications : either the dark energy consists of multiple components with at least one non - canonical phantom component or general relativity needs to be extended to a more general theory on cosmological scales . both of these conjectures have been studied extensively . for instance , curvature quintessence [ 7 ] is a fascinating proposal in this respect . recently , phantom - like behavior in a brane - world model with curvature effects and also in dilatonic brane - world scenario with induced gravity have been reported [ 8 ] . dark energy models with non - minimally coupled scalar field and other extensions of scalar - tensor theories have been studied widely some of which can be found in reference [ 9 ] . in the spirit of scalar - tensor dark energy models , our motivation here is to show that a general @xmath0 dgp - inspired scenario can account for late - time acceleration and crossing of phantom divide line in some suitable domains of model parameters space . to show this feature , first we study cosmological dynamics of @xmath0 dgp - inspired scenario briefly . in the minimal case , motivated by modified theories of gravity , some authors have included a term of the type @xmath6 in the action [ 7,10 ] . this extension for some values of @xmath7 has the capability to explain late - time acceleration of the universe in a simple manner . it is then natural to extend this scenario to more general embedding of dgp inspired scenarios . the purpose of this paper is to perform such a generalization to study both late - time acceleration and possible crossing of phantom divide line in this setup . we focus mainly on the potential of the type @xmath8 for non - minimally coupled scalar field . our setup for minimal case predicts a power - law acceleration supporting observed late - time acceleration . in the non - minimal case , by a suitable choice of non - minimal coupling and scalar field potential , one obtains accelerated expansion in some specific regions of parameters space . while a single minimally coupled scalar field in four dimensions can not reproduce a crossing of the phantom divide line for any scalar field potential [ 11 ] , a non - minimally coupled scalar field account for such a crossing [ 6 ] . in dgp model , equation of state parameter of dark energy never crosses @xmath9 line , and universe eventually turns out to be de sitter phase . nevertheless , in this setup if we include a single scalar field ( ordinary or phantom ) on the brane , we can show that equation of state parameter of dark energy can cross phantom divide line [ 12 ] . crossing of phantom divide line with non - minimally coupled scalar field on the warped dgp brane has been studied recently [ 13 ] . our purpose here is to obtain an extension of these non - minimal dark energy models within dgp - inspired @xmath0 braneworld scenario . we use a prime for differentiation with respect to @xmath10 . an overdot marks differentiation with respect to the brane time coordinate . we start with the following action @xmath11 where the first term shows the usual einstein - hilbert action in 5d bulk with 5d metric denoted by @xmath12 and ricci scalar denoted by @xmath13 . the second term on the right is a generalization of the einstein - hilbert action induced on the brane . this is an extension of the scalar - tensor theories in one side and a generalization of @xmath14-gravity on the other side . we call this model as @xmath0 dgp - inspired scenario . @xmath15 is the coordinate of the fifth dimension and we suppose that brane is located at @xmath16 . @xmath17 is induced metric on the brane which is connected to @xmath12 via @xmath18 . @xmath19 is the trace of the mean extrinsic curvature of the brane defined as follows @xmath20_{y=-\epsilon } + [ k_{\mu\nu}]_{y=+\epsilon}\bigg)\end{aligned}\ ] ] we denote matter field lagrangian by @xmath21 with the following energy - momentum tensor @xmath22 the pure scalar field lagrangian is @xmath23 which gives the following energy - momentum tensor @xmath24 the field equations resulting from this action are given as follows @xmath25 in this relation @xmath26 where @xmath27 is the energy - momentum tensor in matter frame and @xmath28 . also , @xmath29 is defined as follows @xmath30.\ ] ] in the bulk , @xmath31 and therefore @xmath32 and on the brane we have @xmath33 where @xmath34 . the corresponding junction conditions relating quantities on the brane are as follows @xmath35_{y=-\epsilon}^{y=+\epsilon}= \frac{f'(r,\phi)}{m_{4}^{3}}\bigg[\tau_{\mu\nu}-\frac{1}{3}q_{\mu\nu}q^{\alpha\beta } \tau_{\alpha\beta}\bigg]_{y=0}-\]]@xmath36_{y=0}\end{aligned}\ ] ] a detailed study of weak field limit of this scenario within harmonic gauge on the longitudinal coordinates and using green s method to find gravitational potential , leads us to a modified ( effective ) cross - over distance in this set - up as follows ( see [ 14 ] for details of a similar argument ) @xmath37 where @xmath38 the gravitational potential in this scenario takes the following forms in two different extreme : + for @xmath39 @xmath40\end{aligned}\ ] ] and for @xmath41 @xmath42.\end{aligned}\ ] ] @xmath43 is effective mass of the scalar field @xmath44 which is non - minimally coupled to induced gravity , @xmath45 collectively shows the effective mass of other material fields minimally coupled to gravity and @xmath46 is euler s constant . the most important feature of this @xmath0 dgp - inspired scenario is the fact that now cross - over scale is explicitly related to the induced curvature on the brane and non - minimal coupling of scalar field with this curvature . since the dynamics of scalar field @xmath44 is described by @xmath47 , one may explain this result as a spacetime variation of the newton s constant . in this viewpoint , when @xmath0 varies from point to point on dgp brane , the crossover scale takes different values . recent best - fit crossover scale is given by @xmath48 [ 15 ] where @xmath49 . since @xmath50 , if we choose @xmath51 it is possible to constraint this @xmath0 scenario to be consistent with observational data . for instance , if we set @xmath52 it is easy to show that @xmath53 can attain the value of @xmath54 where we have assumed @xmath55 and @xmath56 . this argument has the potential to be used as a mechanism for obtaining reliable form of @xmath0 in cosmological context . embedding of frw cosmology in dgp scenario ( which deviates from general relativity at large distances ) is possible in the sense that this model accounts for cosmological equations of motion at any distance scale on brane with any function of ricci scalar . it is well - known that original dgp model explains late - time acceleration via leakage of gravity to extra dimension . on the other hand , in this model equation of state parameter of dark energy never crosses the phantom divide line . extension of dgp scenario to more generalization of the brane action may result in several new implications on cosmological ground . in this respect , it is interesting to know whether dgp - inspired @xmath57 gravity can account for late - time acceleration and especially phantom divide line crossing . with this motivation , in this section we investigate cosmological implications of our setup focusing firstly on the late - time acceleration . we assume the following line element @xmath58 where @xmath59 is a maximally symmetric 3-dimensional metric defined as @xmath60 where @xmath61 parameterizes the spatial curvature and @xmath62 . we assume that the scalar field @xmath44 depends only on the cosmic time of the brane . choosing a gaussian normal coordinate system so that @xmath63 , non - vanishing components of einstein s tensor in the bulk plus junction conditions on the brane defined as @xmath64_{y=-\epsilon}^{y=+\epsilon}(t)= \frac{2nm_{3}^{2}}{m_{4}^{3}}\bigg [ \big(\frac{df}{dr}\big)\big(\frac{\ddot{a}}{n^{2}a}-\frac{\dot{a}^{2}}{2n^{2}a^{2 } } -\frac{\dot{n}\dot{a}}{n^{3}a}-\frac{k}{2a^{2}}\big)\bigg]_{y=0}\ ] ] @xmath65_{y=0},\end{aligned}\ ] ] @xmath66_{y=-\epsilon}^{y=+\epsilon}(t)= \frac{m_{3}^{2}}{m_{4}^{3}}\bigg[\big(\frac{df}{dr}\big ) \big(\frac{\dot{a}^{2}}{n^{2}a}+\frac{k}{a}\big)\bigg]_{y=0}- \bigg[\bigg(\frac{df}{dr}\bigg)\frac{\rho^{(tot)}a}{3m_{4}^{3}}\bigg]_{y=0}\end{aligned}\ ] ] yield the following generalization of friedmann equation for cosmological dynamics on the brane ( see [ 14 ] for machinery of calculations for a simple case ) @xmath67}\bigg]\bigg)\end{aligned}\ ] ] where @xmath68 shows two different embedding of the brane , @xmath69 and @xmath70 is a constant ( with @xmath71 ) . total energy density and pressure are defined as @xmath72 , @xmath73 respectively . the ordinary matter on the brane has a perfect fluid form with energy density @xmath74 and pressure @xmath75 , while the energy density and pressure corresponding to non - minimally coupled scalar field and also those related to curvature are given as follows @xmath76_{y=0},\end{aligned}\ ] ] @xmath77_{y=0}.\end{aligned}\ ] ] also @xmath78 -3\dot{r}h f''(r,\phi ) \bigg),\end{aligned}\ ] ] @xmath79}\bigg).\end{aligned}\ ] ] where @xmath80 is the hubble parameter . ricci scalar on the brane is given by @xmath81.\ ] ] in this setup , non - minimal coupling of scalar field and induced gravity leads to no - conservation of effective energy density on the brane @xmath82 it is easy to show that for a minimally coupled scalar field on the brane , this setup yields a late - time accelerating universe in a fascinating manner [ 10 ] . the evolution of the scalar field on the brane for a spatially flat frw geometry , @xmath83 , is described by the following equations @xmath84 and @xmath85 and finally @xmath86.\end{aligned}\ ] ] these equations show that essentially embedding of frw cosmology in dgp setup is possible . now , after a brief study of gravitational and cosmological implications of our setup , we investigate late - time acceleration and crossing of the phantom divide line in this setup . standard dgp model itself has the capability to explain the late - time acceleration of the universe via leakage of gravity to extra dimensions without any additional mechanism [ 4 ] . in our @xmath0 dgp inspired model , it is interesting to see whether there is any room for explanation of this late time acceleration . the viability of this question lies in the fact that generally with non - minimal coupling it is harder to achieve accelerated expansion [ 16 ] . we first obtain a necessary condition for the acceleration of the universe in @xmath0 dgp - inspired model . then we use a simple ansatz to clarify our general equations . suppose that scalar field is the only source of matter on the brane so that @xmath87 for other matter candidates . the necessary condition for acceleration of the universe is @xmath88 because in our @xmath0 setup scalar field and curvature are correlated . therefore we obtain @xmath89-\]]@xmath90 using the klein - gordon equation ( 23 ) , this relation can be rewritten as follows @xmath91-\]]@xmath92 this is a general condition for acceleration of the universe in our @xmath0 dgp inspired model . it is a complicated relation and one can not achieve an explicit intuition of acceleration in this setup . for simple forms of @xmath0 we can find relatively simple conditions that can be explained more explicitly . as a simple example and following faraoni [ 16 ] , if we set @xmath93 which gives conformal coupling of scalar field and einstein gravity , equation ( 25 ) under assumption of weak energy condition @xmath94 , gives @xmath95 , where we have assumed @xmath96 [ 16 ] . for instance , if we set @xmath8 , we find @xmath97 . using the ansatz @xmath98 and @xmath99 , equations ( 22 ) and ( 23 ) with @xmath100 and @xmath87 , give @xmath101 with positive and real @xmath102 and considering terms of order @xmath103 . in this case @xmath104 is restricted to the interval @xmath105 and since equation ( 23 ) with these ansatz gives @xmath106 , positivity and reality of solutions for @xmath107 gives @xmath108 . for @xmath109 , we find @xmath110 which gives a power - law accelerated expansion for positive sign . therefore , a suitable fine - tuning of non - minimal coupling provides the possibility of late - time accelerated expansion . based on a dark energy model , to have an accelerated universe , the value of conformal nmc should be restricted to the interval @xmath111 [ 17 ] . on the other hand , current experimental limits on the time variation of @xmath112 constraint the nonminimal coupling as @xmath113[18 ] . solar system experiments such as shapiro time delay and deflection of light have constraint brans - dike parameter to be @xmath114 which leads to the result of @xmath115 for non - minimal coupling [ 18 ] . however , other approaches lead to different constraints on the value of conformal coupling [ 19 ] . as a second and more general example with @xmath116 correction , we set @xmath117 . for spatially flat frw geometry the ricci scalar is given by @xmath118 using the above proposed ansatz for @xmath119 and @xmath120 , equations ( 22 ) and ( 24 ) can be rewritten as the following explicit time - dependent form @xmath121 @xmath122 @xmath123 @xmath124 @xmath125 our aim is to see whether these equations account for positively accelerated expansion with suitable values of parameters . since the parameter space of the problem is complicated , analytical solutions of these equations have no obvious interpretation . therefore , we try to see how a suitable choice of parameters leads to viability of accelerated expansion in this setup . our strategy is to see how with different choices of parameters in equations ( 27 ) and ( 28 ) , equality in this equations is preserved . firstly , we choose @xmath126 in the favor of late - time positively accelerated expansion . we find appropriate values of other parameters in this parameters space as shown in table @xmath127 and @xmath128 . we plot each side of equation ( 27 ) as a separate function of time for some values of parameters . existence of intersection point for these functions shows the viability of values attributed to parameters to preserve equality . the same procedure for both sides of equation ( 28 ) is done too . in summary , to have accelerated expansion , we need @xmath126 , so choosing @xmath126 we find some possible values of the other parameters to have at least one intersection point in the graph and therefore to be consistent with accelerated expansion . secondly , choosing some specific values for other parameters , we obtain possible values of @xmath102 . if there is no intersection between two graphs , it means that with that choice of parameters we will not arrive at @xmath126 corresponding to accelerated expansion . if there is intersection but with @xmath129 , the solution will be decelerating ruled out by observational data . the results of these analysis are shown in table @xmath127 and @xmath128 and also in figures @xmath127 , @xmath128 , @xmath130 and @xmath131 . note that these are only some specific examples and the actual parameter space is very large and complicate . however , the least advantage of this analysis is the fact that essentially late - time acceleration can be explained naturally within our setup . we see for instance that with @xmath132 , @xmath133 , and @xmath134 there is no possibility to have accelerated expansion with @xmath135 since there is no intersection point in the graph . on the other hand , accelerated expansion with @xmath136 requires @xmath137 or @xmath132 , @xmath138 , and @xmath139 for instance . as a common role in this analysis , we have noticed that from equation ( 27 ) , only for positive values of @xmath7 with @xmath140 and for positive or negative @xmath104 there are intersection points and therefore accelerated expansion . these analysis , though very simple and especial , show that @xmath0 dgp - inspired scenarios essentially account for positively accelerated expansion . .some choices of parameters for equation ( 27 ) to be preserved . the last column shows the coordinate of intersection points and equality of two sides of equation ( 27 ) with corresponding choices of parameters . [ cols="^,^,^,^,^,^,^,^",options="header " , ] note that in minimal case where @xmath141 one finds from ( 22 ) and ( 24 ) [ 20 ] @xmath142=0\end{aligned}\ ] ] @xmath143=n(2n-1)(n-1)\end{aligned}\ ] ] from which the allowed solutions are @xmath144@xmath145 our choices of parameters in preceding discussions are based on this argument and also constraints on non - minimal coupling from reference [ 19 ] . especially , we have taken into account the following considerations : the solutions with @xmath146 are not interesting at all since they provide static cosmologies with non - evolving scale factor on the brane . in ordinary 4d gravity with the action of the form @xmath147 and without any matter , a constant scale factor with a singular equation of state is achieved for @xmath148 . similarly , in the normal dgp model , that is , when @xmath148 and with no ordinary matter present , we can not define the equation of state because the scale factor on the brane is constant and equation of state parameter will be singular [ 10,20 ] . in our @xmath0 setup with @xmath149 , as we have discussed previously , to have accelerated expansion we need to rstrict @xmath7 to the interval @xmath140 . in this situation , there is no term with the first power of the ricci scalar to obtain ordinary einstein - like interactions between two astrophysical objects , such as the sun and mercury . to overcome this problem in our setup , we have treated the case with @xmath148 separately after equation ( 25 ) above . on the other hand , if we choose theories such as @xmath150 $ ] where @xmath151 is a suitably chosen parameter , our model will contain the case with @xmath148 and therefore general relativity . we will come back to this issue later with more details . because of explicit interaction between scalar field and curvature , the equation of state parameter for non - minimally coupled scalar field in this @xmath0 dgp - inspired setup is as follows @xmath152 which can be written as @xmath153}{\frac{1}{2}\dot{\phi}^{2}+v(\phi)- 6\frac{d f}{d\phi}h\dot{\phi}+\frac{m_{3}^{2}}{f'(r,\phi)}\big(\frac{1}{2}f(r,\phi)- \frac{1}{2}r f'(r,\phi)-3\dot{r}h\frac{d^2f}{dr^2}\big)},\end{aligned}\ ] ] this is a general statement for equation of state in @xmath0 setup . to proceed further , we have to specify the functional form of @xmath0 . firstly we assume a conformal coupling between scalar field and induced @xmath116-gravity as @xmath154 . the scalar field potential is chosen to be @xmath8 and then using natural ansatz @xmath98 and @xmath155 , the following time - dependent equation of state parameter will be derived @xmath156@xmath157@xmath158 the equation of state parameter in this scenario has a very complicated form . to find further intuition , this equation is plotted for several interesting cases in which follows . as these figures show , essentially crossing of phantom divide line , @xmath159 , is supported in this @xmath0 dgp - inspired model . however there are some points that should be stressed here . as figure @xmath160 shows , the minimal case for a single scalar field has no phantom divide line crossing for some region of parameter space . on the other hand , this non - minimal @xmath0 setup has no crossing of phantom divide line with negative @xmath7 values . however , as table @xmath128 and related arguments in preceding section show , accelerated expansion is possible even with negative @xmath7 values . this is supported by other studies [ 21 ] . for @xmath161 we have a power - low acceleration on the brane without having to introduce dark energy . this result is consistent with the observational results similar to dark energy with the equation of state parameter @xmath162 which has no phantom divide line crossing . figures @xmath163 and @xmath164 show the time evolution of equation of state parameter with some suitable values of model parameters . as we have stressed previously , the case with @xmath148 needs further discussion . we have adopted the ansatz @xmath154 and our analysis has shown that @xmath7 should be restricted to the interval @xmath140 to have reliable cosmology in this setup . in this situation there is no term with the first power of the ricci scalar to obtain ordinary einstein - like interactions between two astrophysical objects , such as the sun and mercury . to overcome this problem in our dgp - inspired @xmath0 gravity we have treated the case with @xmath148 separately in subsection 3.1 and after equation ( 25 ) . form more general viewpoint , we can unify these arguments by choosing the following form of @xmath0 @xmath165,\ ] ] where @xmath151 is a suitably chosen parameter [ 3,21 ] . the accelerated expansion and crossing of phantom divide line in this setup can be studied in the line of previous arguments in this paper . fortunately this type of theories contain @xmath148 and general relativity as a subset . for instance , the equation of state parameter in this setup takes the following form @xmath166@xmath167\bigg\ } \times\]]@xmath168 where @xmath169 , @xmath170 and @xmath171 . figure @xmath172 and @xmath173 show the result of numerical analysis of this model . we see that these types of theories essentially account for crossing of phantom divide line . in this case as figure @xmath173 shows , we have crossing even with negative values of @xmath7 . note that by choosing negative values of @xmath7 , we can treat theories with @xmath174 in this framework . the model proposed by carroll _ et al _ [ 21 ] , lies in this framework . for treating the issue of late time acceleration in this case , friedmann and acceleration equation with this choice of @xmath0 take the following forms @xmath175@xmath176 and @xmath177@xmath178@xmath179 where @xmath180 , @xmath181 and @xmath182 are defined previously . numerical analysis of these equations shows the possibility of intersection points and therefore accelerated expansion . for instance , figure @xmath183 shows the situation for a specific choice of the parameters . as another important point , the issue of stability of the self - accelerated solutions should be stressed here . the self - accelerating branch of the dgp model contains a ghost at the linearized level [ 22 ] . since the ghost carries negative energy density , it leads to the instability of the spacetime . the presence of the ghost can be attributed to the infinite volume of the extra - dimension in dgp setup . when there are ghosts instabilities in self - accelerating branch , it is natural to ask what are the results of solutions decay . as a possible answer we can state that since the normal branch solutions are ghost - free , one can think that the self - accelerating solutions may decay into the normal branch solutions . in fact for a given brane tension , the hubble parameter in the self - accelerating universe is larger than that of the normal branch solutions . then it is possible to have nucleation of bubbles of the normal branch in the environment of the self - accelerating branch solution . this is similar to the false vacuum decay in de sitter space . however , there are arguments against this kind of reasoning which suggest that the self - accelerating branch does not decay into the normal branch by forming normal branch bubbles ( see [ 22 ] for more details ) . it was also shown that the introduction of gauss - bonnet term in the bulk does not help to overcome this problem [ 23 ] . in fact , it is still unclear what is the end state of the ghost instability in self - accelerated branch of dgp inspired setups ( for more details see [ 22 ] ) . on the other hand , non - minimal coupling of scalar field and induced gravity in our @xmath0 setup provides a new degree of freedom which requires special fine tuning and this my provide a suitable basis to treat ghost instability . it seems that in our model this additional degree of freedom has the capability to provide the background for a more reliable solution to ghost instability due to wider parameter space . finally , the phantom divide line crossing in conventional scalar field theory violates positive energy theorems . in our setup , non - minimal coupling of scalar field and induced gravity has the capability to evade this problem . in fact , following lue and starkman [ 24 ] , if we consider the frw phase instead of the self - accelerating phase , and relax the presumption that the cosmological constant be zero ( i.e. , abandon the notion of completely replacing dark energy ) , then we can achieve @xmath184 without violating the null - energy condition , without ghost instabilities and without a big rip ( see [ 24 - 27 ] for more details ) . current positively accelerated expansion of the universe could be the result of a modification to the einstein - hilbert action . also dgp braneworld scenario has the capability to interpret this late - time acceleration via leakage of gravity to extra dimension in its self - accelerating branch . on the other hand , there are several compelling reasons for inclusion of an explicit non - minimal coupling of scalar field and induced gravity on the brane sector of dgp action [ 14,16 ] . these arguments have led us to study cosmological implications of a dgp - inspired @xmath0 gravity . although this issue has been studied extensively for the minimal case , our setup provides a new approach to treat accelerated expansion within a general framework of _ modified _ scalar - tensor theories taking into account the role played by curvature correction and non - minimal coupling of scalar field and modified curvature simultaneously . we have studied late - time acceleration and possible crossing of phantom divide line in this setup . the condition for accelerated expansion and also equation governing on the dynamics of equation of state parameter are very complicated , so that we were forced to try a reliable and natural ansatz to find some intuition . we have shown that this setup accounts for accelerated expansion with a suitable choice of parameters space or fine - tuning . also , this setup accounts for crossing of phantom divide line in some ranges of parameters appeared in the model . in the minimal case , our model coincides with existing models of @xmath14 gravity . especially the case with @xmath161 in the absence of non - minimal coupling exactly coincides with well - known result [ 21 ] . on the other hand , for a single scalar field with minimal coupling to gravity there are region of parameter space with no phantom divide line crossing . we have seen in the numerical calculations that with @xmath185 , for both @xmath137 and @xmath132 and with arbitrary @xmath104 , there is no crossing of phantom divide line for negative values of @xmath7 ( figure @xmath163 ) . however positively accelerated expansion of the universe can be explained in this situation with negative @xmath7 . on the other hand , if @xmath140 , for negative @xmath104 we have crossing of the phantom divide line ( see figure @xmath164 ) . as an important result in the analysis of late - time behavior , from equation ( 27 ) we find that only for positive values of @xmath7 with @xmath140 and for positive or negative values of @xmath104 , there are intersection points and therefore accelerated expansion for corresponding parameters choice . the case with @xmath148 which describes ordinary general relativity has been treated separately . by adopting more general theories such as @xmath150 $ ] , the previous restriction on @xmath7 is relaxed so that @xmath148 is a natural subset of the model . in summary , accelerated expansion can be explained by both positive and negative values of @xmath7 with suitable fine - tuning of other parameters . on the other hand crossing of phantom divide line in @xmath117 model can be explained only with positive values of @xmath7 and fine - tuning of other parameters . the range of variation of @xmath7 to have crossing of phantom divide line in this setup is @xmath140 . however in theories with @xmath150 $ ] this crossing occurs for both negative and positive values of @xmath7 . for theories of the type @xmath150 $ ] , @xmath7 is not restricted to this interval and these theories account for phantom divide line crossing even for negative values of @xmath7 as figure @xmath173 shows . the issues of ghost instabilities and violation of positive energy theorems have been discussed also . due to wider parameter space as a result of non - minimal coupling of scalar field and modified gravity , we hope this model can evade these problems . + * acknowledgment * we are indebted to an anonymous referee for his / her important contribution in this work . 10 a. g. riess _ et . ( supernova search team collaboration ) , astron . j. * 116 * ( 1998 ) 1006 , [ astro - ph/9805201 ] ; s. perlmutter _ et . al . _ , astrophys . j. * 517 * ( 1999 ) 565 , [ astro - ph/9812133 ] ; d. n. spergel _ et . * 148 * ( 2003 ) 175 , [ astro - ph/0302209 ] ; c. l. bennett _ et . * 148 * ( 2003 ) 1 , [ astro - ph/0302207 ] ; c. b. netterfield _ et . j. * 571 * ( 2002 ) 604 , [ astro - ph/0104460 ] ; n. w. halverson _ et . j. * 568 * ( 2002 ) 38 , [ astro - ph/0104489 ] ; d. n. spergel _ et al _ , _ wmap three year results : implication for cosmology _ , [ arxiv : astro - ph/0603499 ] . e. j. copeland , m. sami and s. tsujikawa , int . d * 15 * ( 2006 ) 1753 - 1936 , [ arxiv : hep - th/0603057 ] . t. p. sotiriou and v. faraoni , [ arxiv:0805.1726 ] . c. deffayet , g. dvali and g. gabadadze , phys . d * 65 * ( 2002 ) 044023 , [ arxiv : astro - ph/0105068 ] . t. padmanabhan , phys . rept . * 380 * ( 2003 ) 253 ; v. shani and a. a. astarobinsky , int . d * 9 * ( 2000 ) 373 ; s. m. carroll , living rev . rel . * 4 * ( 2001 ) 1 . s. nesseris and l. perivolaropoulos , jcap * 0701 * ( 2007 ) 018 ; [ arxiv : astro - ph/0610092 ] . s. capozziello , v. f. cardone and a. troisi , jcap * 0608 * ( 2006 ) 001 , [ arxiv : astro - ph/0602349 ] ; s. capozziello , s. nojiri and s.d . odintsov , phys . b * 634 * ( 2006 ) 93 , [ arxiv : hep - th/0512118 ] ; s. capozziello , v. f. cardone , e. piedipalumbo and c. rubano , class . * 23 * ( 2006 ) 1205 , [ arxiv : astro - ph/0507438 ] ; s. capozziello , v. f. cardone , s. carloni and a. troisi , int phys . d12 ( 2003 ) 1969 , [ arxiv : astro - ph/0307018 ] . m. bouhmadi - lopez and p. vargas moniz , [ arxiv:0804.4484 ] ; m. bouhmadi - lopez , [ arxiv : astro - ph/0512124 ] . r. gannouji , d. polarski , a. ranquet and a. a. starobinsky , jcap * 0609 * ( 2006 ) 016 , [ arxiv : astro - ph/0606287 ] ; o. hrycyna and m. szydlowski , phys . lett . b651 ( 2007 ) 8 , [ arxiv:0704.1651 ] ; o. hrycyna and m. szydlowski , phys . rev . d76 ( 2007 ) 123510 , [ arxiv:0707.4471 ] ; m. szydlowski , o. hrycyna and a. kurek , phys . rev . d * 77 * ( 2008 ) 027302 , [ arxiv:0710.0366 ] ; s. nojiri , s. d. odintsov and p. v. tretyakov , [ arxiv:0710.5232 ] k. atazadeh , m. farhoudi and h. r. sepangi , phys . b * 660 * ( 2008 ) 275 , [ arxiv:0801.1398 ] ; k. atazadeh and h. r. sepangi , jcap * 09 * ( 2007 ) 020 , [ arxiv:0710.0214 ] . a. vikman , phys . d * 71 * ( 2005 ) 023515 , [ astro - ph/0407107 ] . see also other papers which accepted almost this viewpoint : y. h. wei and y. z. zhang , grav . * 9 * ( 2003 ) 307 ; v. sahni and y. shtanov , jcap * 0311 * ( 2003 ) 014 ; y.h . wei and y. tian , class . quantum grav . * 21 * ( 2004 ) 5347 ; f. c. carvalho and a. saa , phys . d * 70 * ( 2004 ) 087302 ; f. piazza and s. tsujikawa , jcap * 0407 * ( 2004 ) 004 ; r - g . cai , h. s. zhang and a. wang , commun . 44(2005 ) 948 ; i. y. arefeva , a. s. koshelev and s. y. vernov , phys . d 72 * ( 2005 ) 064017 ; a. anisimov , e. babichev and a. vikman , jcap * 0506 * ( 2005)006 ; b. wang , y.g . gong and e. abdalla , phys . b * 624 * ( 2005 ) 141 ; s. nojiri and s. d. odintsov , hep - th/0506212 ; s. nojiri , s. d. odintsov and s. tsujikawa , hep - th/0501025 ; e. elizalde , s. nojiri , s. d. odintsov and p. wang , phys . d * 71 * ( 2005 ) 103504 ; h. mohseni sadjadi , phys . rev . d * 73 * ( 2006 ) 063525 ; w. zhao and y. zhang , phys . rev.d * 73 * ( 2006)123509 ; p. s. apostolopoulos and n. tetradis , hep - th/0604014 ; i. ya . arefeva and a. s. koshelev , hep - th/0605085 ; s. d. sadatian and k. nozari , europhys . lett . * 82 * ( 2008 ) 49001 . h. zhang and z .- h . zhu , phys . d * 75 * ( 2007 ) 023510 , [ arxiv : astro - ph/0611834 ] k. nozari , n. behrouz and b. fazlpour , [ arxiv:0808.0318 ] k. nozari , jcap * 09 * ( 2007 ) 003 , [ arxiv : hep - th/07081611 ] . et al . _ , j. * 607 * ( 2004 ) 665 . v. faraoni , phys . d * 62 * ( 2000 ) 023504 , [ arxiv : gr - qc/0002091 ] ; v. faraoni , _ phys . * 53 * ( 1996 ) 6813 . m. ito , europhys . * 71 * ( 5 ) ( 2005 ) 712 t. chiba , phys . d * 60 * ( 1999 ) 083508 k. nozari and s. d. sadatian , [ arxiv:0710.0058 ] , to appear in mod . s. capozziello , int . phys . d11 ( 2002 ) 483 , [ arxiv : gr - qc/0201033 ] ; s. carloni , p. k. s. dunsby , s. capozziello and a. troisi , class . * 22 * ( 2005 ) 4839 , [ arxiv : gr - qc/0410046 ] . s. m. carroll , v. duvvuri , m. trodden , m. turner , phys . d * 70 * ( 2004 ) 043528 , [ astro - ph/0306438 ] . k. koyama , _ class . quantum grav . _ * 24 * , r231 ( 2007 ) [ arxiv : hep - th/0709.2399 ] . c. de rham and a. j. tolley , _ jcap _ * 0607 * , 004 ( 2006 ) [ arxiv : hep - th/0605122 ] . v. sahni and y. shtanov , _ jcap _ * 0311 * ( 2003 ) 014 , [ arxiv : astro - ph/0202346 ] u. alam and v. sahni , [ arxiv : astro - ph/0209443 ] . a. lue and g. d. starkman , _ phys . d _ * 70 * ( 2004 ) 101501 , [ arxiv : astro - ph/0408246 ] . a. lue , _ phys . * 423 * ( 2006 ) 1 , [ arxiv : astro - ph/0510068 ] .	we study possible crossing of the phantom divide line in a dgp - inspired @xmath0 braneworld scenario where scalar field and curvature quintessence are treated in a unified framework . with some specific form of @xmath0 and by adopting a suitable ansatz , we show that there are appropriate regions of the parameters space which account for late - time acceleration and admit crossing of the phantom divide line . * key words * : braneworld cosmology , dgp scenario , dark energy models , late - time acceleration + * pacs * : 04.50.+h , 98.80.-k 0 mm 0 mm 0 mm 0 mm 0 mm 165 mm .65 cm + and +
in this paper we consider @xmath0 positive masses in an euclidean space @xmath1 , submitted to a gravitational interaction . we find some interesting solutions with a given asymptotic behaviour . the equation of motion of the @xmath0-body problem is written @xmath5 where @xmath6 is the mass and @xmath7 the position of the @xmath8-th body . since these equations are invariant by translation , we can assume that the center of mass is at the origin . + these equations are euler - lagrange equations of the lagrangian action functional ( we will define it precisely in the next section ) , therefore solutions of ( [ newt ] ) are critical points of the action in a set of paths with fixed ends . the simplest kind of critical points are minima , so it is natural to search for minimizers of the lagrangian action joining two given configurations in a fixed time . the potential of the n - body problem being singular at collision configurations , a main difficult involved in this approach is to show that minimizers are collision - free . the following theorem , essentially due to c. marchal , is a major advanced in this subject . given two n - body configurations @xmath9 , @xmath10 and a time @xmath11 , an action minimizing path joining @xmath3 to @xmath12 in time @xmath13 is collision - free for @xmath14 . see [ ma1 ] , [ ma2 ] and [ ch2 ] for a claim and a proof of this theorem for @xmath15 and @xmath16 . see [ fe - te ] for a proof in any dimension . this theorem , together with the lower semicontinuity of the action ( see section [ variat - setting ] ) , implies in particular that there always exists a collision - free minimizing solution joining two given collision - free n - body configuration in a given time . + a natural extension of marchal s theorem is to search solutions defined on an infinite interval @xmath17 , starting from a given configuration at @xmath18 and having a given asymptotic behaviour for @xmath19 . the classification of all possible asymptotic behaviour of solutions in the n - body problem has been investigated since the beginning of the last century . the main results in this direction are due to j. chazy . in [ cha1 ] it is shown that there are only seven possible final evolutions in the three - body problem . among these seven possibilities there are the so - called _ parabolic motions_. a solution @xmath20 of the n - body problem is said to be parabolic if the velocity of every body tends to zero as @xmath19 . we introduce the functions @xmath21 respectively equal to the moment of inertia with respect to the center of mass and to the newtonian potential . given a configuration @xmath22 , we denote by @xmath23 the associated normalized configuration . it is well known ( see for istance [ hu - sa ] and [ ch1 ] ) that if @xmath24 is a parabolic solution , the normalized trajectory @xmath25 is asymptotic to the set of _ central configurations _ ( i.e. critical points of @xmath26 ) . given a central configuration @xmath4 with @xmath27 , we say that a parabolic solution @xmath24 is asymptotic to @xmath4 if @xmath28 as @xmath19 . a central configuration @xmath4 is said to be minimizing if it is an absolute minimum of @xmath29 . we can now state the main result of this paper . given any initial configuration @xmath3 and any minimizing normalized central configuration @xmath4 , there exists a parabolic solution @xmath30 starting from @xmath3 at @xmath18 and asymptotic to @xmath4 for @xmath19 . this solution is a minimizer of the lagrangian action with fixed ends in every compact interval contained in @xmath17 and it is collision - free for @xmath31 . we do not require any hypothesis of nondegeneracy of the central configuration @xmath4 . + the parabolic solution @xmath32 is constructed as limit of a sequence @xmath33\rightarrow ( { { \mathbb r}}^d)^n$ ] of minimizers connecting @xmath3 with a configuration homothetic to @xmath4 in time @xmath34 , and @xmath35 . in section [ construct - solut ] we construct the sequence @xmath36 and we prove that it is uniformly convergent on every compact subset of @xmath37 . in sections [ parabol - solut ] and [ lambert ] we show that @xmath32 is parabolic and asymptotic to @xmath4 . the proof of this last property is achieved by comparing the action of the n - body problem with the action of a kepler problem , and using lambert s theorem to estimates the action . in the appendix we state and prove some technical estimates concerning the kepler problem on the line that we need to construct @xmath32 and to prove its parabolicity . the authors believe that these minimizing parabolic solutions are in fact calibrated curves of some weak kam solutions of the n - body problems , whose existence has been proved in [ mad ] by one of the authors . our main theorem has a natural interpretation in terms of mcgehee vector field and collision manifold . indeed , in [ ch1 ] , [ mcg ] and [ mo ] it is shown that if @xmath4 is a central configuration with @xmath27 , the state @xmath38 with @xmath39 is a critical point of the mcgehee vector field in the collision manifold , and its stable set corresponds to parabolic solutions asymptotic to @xmath4 as @xmath19 . thus , we can formulate the main theorem by saying that the stable set of @xmath38 ( for the mcgehee vector field ) projects on the whole configuration space , provided @xmath4 is a minimizing central configuration . + we think that variational methods could be used to study some important features on the global dynamics of n - body problem . in particular , it should be interesting to study hyperbolic solutions using variational methods . we recall that a solution @xmath40 is said to be hyperbolic if there exists a ( collision - free ) configuration @xmath4 such that @xmath41 a hyperbolic solution has necessarily positive energy , and replacing @xmath4 by a normalized configuration , ( [ solut - hyperb ] ) is equivalent to @xmath42 as @xmath19 ( see [ cha1 ] ) , where @xmath43 is the energy of the solution . in this case we will say that @xmath44 is hyperbolic for @xmath19 and asymptotic to @xmath4 . since there is no constraint to the limit configuration @xmath4 of a hyperbolic solution ( see again [ cha1 ] ) , it is natural to ask the following two questions . the second one has been asked by r. montgomery . + * question 1 . * given an initial configuration @xmath3 and a normalized non - collision configuration @xmath4 , does there exist a hyperbolic motion starting from @xmath3 at @xmath18 and asymptotic to @xmath4 for @xmath19 ? + * question 2 . * for which couple of normalized non - collision configurations @xmath4 and @xmath45 does there exist a solution that is hyperbolic both for @xmath19 and for @xmath46 and is asymptotic to @xmath4 for @xmath19 and to @xmath45 for @xmath46 ? + we hope that it will be possible to answer these questions using variational methods similar to those developed in this paper . since equations ( [ newt ] ) are invariant by translation , we fix the origin of our inertial frame at the center of mass of the system . we define the _ configuration space _ of the system as @xmath47 and we endow @xmath48 with the _ mass scalar product _ : @xmath49 where @xmath50 is the usual euclidean product in @xmath1 . we denote by @xmath51 the euclidean norm on @xmath48 associated to the mass scalar product . a configuration @xmath52 is said to be a _ collision configuration _ if @xmath53 for some @xmath54 . we denote by @xmath55 the set of collision configurations and by @xmath56 the set of collisions - free configurations . equations ( [ newt ] ) can be written in a more compact form as a second order differential equation on @xmath57 @xmath58 where @xmath59 is the newtonian potential already defined in ( [ potentiel&moment - d - inertie ] ) , the gradient is calculated with respect to the mass scalar product . since @xmath57 is an open subset of @xmath48 , the tangent space of @xmath57 is identified with @xmath60 . the following functions defined on @xmath60 @xmath61 are respectively equal to twice the kinetic energy , to the lagrangian and to the energy first integral . + given an absolutely continuous path @xmath62\rightarrow { { \mathcal x}}$ ] , we define its _ lagrange action _ by : @xmath63 where @xmath64 is naturally extended to a function defined over @xmath65 by @xmath66 if @xmath67 . it is well known that collision - free extremals of @xmath68 are solutions of equations ( [ newt-2 ] ) . we say that an absolutely continuous path @xmath62\rightarrow { { \mathcal x}}$ ] is a minimizer if @xmath69 for every absolutely continuous path @xmath70\rightarrow { { \mathcal x}}$ ] having the same extremities . if @xmath71 is any interval , we say that @xmath72 is minimizing if for every compact interval @xmath73 $ ] contained in @xmath74 , the path @xmath75}\right.$ ] is a minimizer . given a positive real number @xmath13 and two configurations @xmath3 and @xmath12 , let @xmath76 be the set of absolutely continuous paths defined in the interval @xmath77 $ ] and joining @xmath3 to @xmath12 in time @xmath13 . the following proposition is well known . for every @xmath78 and for every @xmath11 there exists a minimizer @xmath79\rightarrow { { \mathcal x}}$ ] joining @xmath3 to @xmath12 . in [ ve ] and [ fe - te ] one can find a proof of this proposition when the functional @xmath68 is defined over @xmath80 paths ( i.e. absolutely continuous paths with derivative in @xmath81 . ) joining @xmath3 to @xmath12 . an absolutely continuous path having a finite action is necessarily in @xmath80 , therefore minimizers among @xmath80 paths are also minimizers among absolutely continuous paths . + the proposition above do not ensure that @xmath32 is collision - free , but by the already cited marchal s theorem , if @xmath82 , minimizers are collision - free for @xmath14 . in this section we construct the solution @xmath83 of the main theorem as limit of minimizers . we will show in sections [ parabol - solut ] and [ lambert ] that @xmath32 is parabolic and asymptotic to @xmath4 . + before stating the main result of this section , we recall a classical result concerning parabolic solutions ( see [ ch1 ] or [ hu - sa ] ) for a proof ) . [ parabolic_sundman ] if @xmath83 is a parabolic solution of the n - body problem , the energy of @xmath32 is necessarily zero , moreover we have @xmath84 as @xmath19 , where @xmath85 in particular , the @xmath86-limit of @xmath87 is contained in the set of normalized central configuration . since there are always infinitely many normalized central configurations for a given critical level of @xmath29 , ( the orthogonal group acts on @xmath57 leaving invariant @xmath29 ) , we can not say _ a priori _ that the @xmath86-limit of @xmath87 is a given configuration . if @xmath44 is a parabolic solution asymptotic to normalized central configuration @xmath4 ( i.e. @xmath87 converges to @xmath4 ) , by proposition [ parabolic_sundman ] we have the asymptotic estimates @xmath88 the following lemma is a converse of proposition [ parabolic_sundman ] . [ lem - parab - asympt - x0 ] let @xmath4 be a normalized central configuration , @xmath89 and @xmath90 the constant defined in ( [ defit - alpha ] ) . a solution @xmath91 satisfying the asymptotic estimates ( [ gamma - i_0-t43 ] ) is parabolic and asymptotic to @xmath4 . * proof . * we just need to prove that @xmath32 is parabolic . replacing ( [ gamma - i_0-t43 ] ) in the equation of motion we find @xmath92 , as @xmath19 . therefore , the velocity @xmath93 has a limit for @xmath19 that we denote @xmath94 . moreover we have @xmath95 integrating this expression we find @xmath96 thus @xmath97 and @xmath44 is parabolic . + by the way , if @xmath4 is a normalized central configuration , the path @xmath98 is a solution of the @xmath0-body problem . @xmath99 is called _ homothetic - parabolic solution asymptotic to _ @xmath4 . + we state now the main result of this section . we recall that @xmath3 is the initial configuration of the main theorem , @xmath4 is a normalized minimizing central configuration , @xmath100 and @xmath90 are as before , @xmath101 is given by ( [ pauline ] ) . [ thm - exist - solut ] there exists a minimizing solution @xmath83 starting from @xmath3 , a sequence of positive numbers @xmath35 and a sequence of minimizers @xmath102 such that @xmath36 converges uniformly to @xmath32 on every compact interval contained in @xmath17 . moreover @xmath44 is collision - free for @xmath31 . we prove this theorem in several steps . at proposition [ lacapelle - marival ] we show that if @xmath13 and @xmath103 are sufficiently great , for every minimizer @xmath104 the action @xmath105}\right.)$ ] has a uniform bound ( independent of @xmath106 ) . successively , using ascoli s theorem and a diagonal trick , we find the sequence @xmath107 . we start with some preliminary definitions and remarks . given two configurations @xmath22 and @xmath108 and a time @xmath13 , we denote by @xmath109 the action of a minimizing path joining @xmath22 to @xmath108 in time @xmath13 ( the same function is denoted @xmath110 in [ mad ] ) . in a similar way , given two positive real numbers @xmath111 and @xmath112 and a time @xmath13 , we denote by @xmath113 the action ( for the one dimensional keplerian problem with lagrangian @xmath114 ) of a minimizing path joining @xmath111 to @xmath112 in time @xmath13 . + by the homogeneity of @xmath59 , if @xmath115\rightarrow { { \mathcal x}}$ ] is a solution of ( [ newt-2 ] ) and @xmath116 , the path @xmath117\rightarrow { { \mathcal x}},\qquad \varpi^\lambda(t)=\lambda^{\frac{2}{3}}\varpi(t/\lambda)\ ] ] is still a solution of ( [ newt-2 ] ) . moreover , if @xmath118 is a minimizer , @xmath119 is still a minimizer . a similar property holds for solutions and minimizers of a one dimensional kepler problem . therefore we have @xmath120 [ nonna - isabella ] we have @xmath121 with equality if and only if @xmath22 and @xmath108 are on the half - line starting from zero generated by @xmath122 , where @xmath122 is a normalized minimizing configuration ( i.e. @xmath123 and @xmath124 ) . * let @xmath115\rightarrow { { \mathcal x}}$ ] be a minimizer joining @xmath22 to @xmath108 in time @xmath13 and let @xmath125 . by sundman inequality we have @xmath126 with equality if and only if @xmath127 is parallel to @xmath128 . since @xmath100 is the minimum of @xmath29 we have also @xmath129 with equality if and only if @xmath130 . therefore @xmath131 with equality if and only if @xmath132 , where @xmath122 is a minimizing normalized configuration and @xmath133 is a minimizer ( for the one - dimensional kepler problem ) joining @xmath134 to @xmath135 in time @xmath13 . this proves the lemma . + in order to simplify the exposition we introduce the following notation . if @xmath136 are two configurations and @xmath137 we term @xmath138 in a similar way , if @xmath139 and @xmath137 we term @xmath140 [ lemme - fondamentale ] let @xmath141 be real numbers . if @xmath142 and @xmath143 we have @xmath144 * proof.to prove the first inequality , let @xmath145 and let @xmath146 . the path @xmath147 is nothing but a repametrization of @xmath148}\right.$ ] . since @xmath149 and @xmath150 are minimizers , we have the triangular inequalities @xmath151}\right . ) \\ a_l(\overline{\gamma } ) & \le a_l(\xi)+a_l(\zeta ) , \end{array}\ ] ] therefore @xmath152}\right.)\le a_l(\xi)+a_l(\overline{\gamma } ) \le 2 a_l(\xi)+a_l(\zeta).\ ] ] this gives the first inequality . + the second inequality is a direct consequence of the first one and of lemma [ nonna - isabella ] . , @xmath149 , @xmath147 and @xmath153 in the configuration space . ] [ lacapelle - marival ] there exist three constants @xmath154 , @xmath155 and @xmath156 such that for every @xmath157 , for every @xmath158 and for every @xmath159 we have @xmath160 proof . * suppose , for the sake of a contradiction , that there exist three sequences of positive real numbers @xmath161 , @xmath162 and @xmath163 satisfying @xmath164 and a sequence of minimizers @xmath165 such that for every @xmath166 : @xmath167 let @xmath168 and @xmath169\rightarrow { { \mathcal x}}$ ] be a minimizer connecting @xmath170 to @xmath3 in time @xmath171 . without loss of generality we can assume @xmath172 . by homothety invariance and by the second inequality of lemma [ lemme - fondamentale ] we have @xmath173 since @xmath174 and @xmath175 , by proposition [ elenuccia ] of the appendix we have @xmath176 as @xmath177 . this contradicts inequality ( [ maria - rosa ] ) . + we need now an estimates of the minimal action @xmath109 when @xmath134 and @xmath135 are less then a given size . [ cap - de - l - homy ] there exist two positive constants @xmath178 and @xmath179 such that if @xmath180 and @xmath11 , if @xmath181 and @xmath182 are two configurations satisfying @xmath183 and @xmath184 , we can find an absolutely continuous path @xmath185\rightarrow { { \mathcal x}}$ ] joining @xmath22 to @xmath108 in time @xmath13 such that the following inequality holds @xmath186 in particular we have @xmath187 * proof.let @xmath188 be any normalized collision - free configuration . we construct an absolutely continuous path @xmath189\rightarrow { { \mathcal x}}$ ] joining @xmath22 to @xmath190 and verifying @xmath191 where @xmath192 and @xmath193 are two positive constants independent on @xmath194 , @xmath13 and @xmath22 . an analogous path @xmath195\rightarrow { { \mathcal x}}$ ] joining @xmath190 to @xmath108 can be constructed in exactly the same way . pasting @xmath196 and @xmath197 together and choosing @xmath198 and @xmath199 we get a path @xmath200 verifying ( [ ezequiel ] ) . inequality ( [ budoni ] ) is an obvious consequence of ( [ ezequiel ] ) . + let @xmath201 . we term @xmath202 and @xmath203 for @xmath204 . in a similar way , given @xmath52 with @xmath183 , we term @xmath205 and @xmath206 . let @xmath207 be the coefficients @xmath208 and let @xmath43 be the cardinality of the set @xmath209 . the inequality @xmath210 holds . let us denote @xmath211 the elements of the set @xmath209 ordered increasingly . we define @xmath212 and @xmath213 . for every @xmath214 we term @xmath215 we observe that @xmath216 and @xmath217 if @xmath218 , moreover @xmath219 . defining @xmath220 we have @xmath221 . let @xmath222\rightarrow [ 0,1]$ ] be the path defined by @xmath223 & \\ \mu_i+\left(\frac{t-\sigma_i}{\tau_i/2}\right)^\frac{2}{3 } \frac{\mu_{i+1}-\mu_i}{2 } , \quad & \text{if}\quad t\in [ \sigma_i,\sigma_i+\frac{\tau_i}{2}],\quad & i=1, ... ,h-1 \\ \mu_{i+1}- \left(\frac{\sigma_{i+1}-t}{\tau_i/2}\right)^\frac{2}{3 } \frac{\mu_{i+1}-\mu_i}{2 } , \quad & \text{if}\quad t\in [ \sigma_i+\frac{\tau_i}{2},\sigma_{i+1 } ] , \quad & i=1, ... ,h-1 \\ \mu_h+(1-\mu_h)\left ( \frac{t-\sigma_h}{t/2-\sigma_h}\right)^\frac{2}{3 } , \quad & \text{if } \quad t\in [ \sigma_h , t/2 ] . & \end{array } \right.\ ] ] the definition of @xmath224 in the interval @xmath225 $ ] has some meaning only if @xmath226 ( i.e. if @xmath227 ) . the path @xmath228\ ] ] connects @xmath22 to @xmath190 in the time @xmath229 . if @xmath230 , the action of the restriction @xmath231}\right.$ ] is given by @xmath232}\right.)&=\frac{\\|r x_0^\prime -x\\|^2}{2 } \displaystyle\int_{\sigma_i}^{\sigma_i+\frac{\tau_i}{2 } } \dot{\lambda}(t)^2\ , dt \\ & + \sum\limits_{1\le j < k\le n } m_j m_k \displaystyle\int_{\sigma_i}^{\sigma_i+\frac{\tau_i}{2 } } \frac{dt}{\|(1-\lambda(t))\vec{r}_{jk}+\lambda(t)r \vec{c}_{jk}\|}. \end{array}\ ] ] as @xmath233 $ ] the path @xmath224 increases from @xmath234 to @xmath235 , hence the coefficient @xmath236 that is closest to @xmath224 is exactly @xmath234 . using the triangular inequality we find @xmath237 for @xmath233 $ ] and for every @xmath238 . therefore , since @xmath183 and @xmath239 @xmath232}\right.)&\le 2r^2 \left(\frac{\mu_{i+1}-\mu_i}{2}\right)^2 \left(\frac{2}{\tau_i}\right)^\frac{4}{3 } \displaystyle\int_{\sigma_i}^{\sigma_i+\frac{\tau_i}{2 } } \frac{4}{9}(t-\sigma_i)^{-\frac{2}{3 } } dt \\ & + \sum\limits_{1\le j < k\le n}\frac{m_j m_k}{rc_{jk } } \displaystyle\int_{\sigma_i}^{\sigma_i+\frac{\tau_i}{2}}\frac{dt}{\lambda(t)-\mu_i } \\ & = \frac{4r^2}{3 } \frac{(\mu_{i+1}-\mu_i)^2}{\tau_i}+\frac{u(x_0^\prime)}{r}\frac{2}{\mu_{i+1}-\mu_i } \left(\frac{\tau_i}{2}\right)^\frac{2}{3 } \displaystyle\int_{\sigma_i}^{\sigma_i+\frac{\tau_i}{2 } } \frac{dt}{(t-\sigma_i)^\frac{2}{3 } } \\ & = \frac{4r^2}{3 } \frac{(\mu_{i+1}-\mu_i)^2}{\tau_i}+\frac{3 u(x_0^\prime)}{r } \frac{\tau_i}{\mu_{i+1}-\mu_i}. \end{array}\ ] ] in a similar way we find @xmath240}\right.)&\le \frac{4r^2}{3 } \frac{(\mu_{i+1}-\mu_i)^2}{\tau_i}+\frac{3 u(x_0^\prime)}{r } \frac{\tau_i}{\mu_{i+1}-\mu_i } , \qquad i=1, ... ,h-1 \\ a_l(\gamma_x\left\|_{[\sigma_0,\sigma_1]}\right.)&\le \frac{8r^2}{3 } \frac{\mu_1 ^ 2}{\tau_0}+\frac{3 u(x_0^\prime)}{r } \frac{\tau_0}{\mu_1 } , \\ a_l(\gamma_x\left\|_{[\sigma_h,\sigma_{h+1}]}\right.)&\le \frac{8r^2}{3 } \frac{(1-\mu_h)^2}{\tau_h}+\frac{3 u(x_0^\prime)}{r } \frac{\tau_h}{1-\mu_h}. \end{array}\ ] ] that gives @xmath241}\right.)\le \frac{8r^2}{3 } \frac{(\mu_{i+1}-\mu_i)^2}{\tau_i}+\frac{6 u(x_0^\prime)}{r } \frac{\tau_i}{\mu_{i+1}-\mu_i } , \qquad i=0, ... ,h,\ ] ] and by definition of @xmath242 @xmath243 by definition of @xmath234 we have @xmath244 let us introduce now the functions @xmath245 and study minima and maxima of @xmath246 and @xmath247 under the condition @xmath248 . we show by induction on @xmath43 that @xmath249 if @xmath250 , condition @xmath248 implies @xmath251 , thus @xmath252 assuming now the statement is true up to order @xmath253 , let us prove it is true at order @xmath43 . by lagrange multiplier theorem , the unique interior critical point of @xmath246 under the condition @xmath248 is given by the equations @xmath254 this gives @xmath255 the boundary of the simplex @xmath248 is the set of @xmath256 such that @xmath257 and @xmath258 for at least one indices @xmath8 . by inductive hypothesis , the minimum of @xmath259 on the boundary of @xmath248 is @xmath260 and the maximum is @xmath261 . comparing with the value of @xmath246 on the unique interior critical point of @xmath246 we find ( [ constraints ] ) . in a similar way one prove @xmath262 replacing these estimates in ( [ act - gamma - x ] ) we find @xmath263 since @xmath264 , inequality ( [ pma ] ) is proved . + we give now the proof of theorem [ thm - exist - solut ] . + proof of theorem [ thm - exist - solut]. by propositions [ lacapelle - marival ] and [ cap - de - l - homy ] , there exist three constants @xmath265 , @xmath266 and @xmath156 such that for every @xmath157 , for every @xmath267 and for every minimizer @xmath143 we have @xmath268}\right.)\le a t^\frac{1}{3}\ ] ] let us prove the equicontinuity of the family @xmath269}\right.,\quad \overline{\gamma}\in \sigma(x_i,\gamma_0(t);t ) , \quad t\ge \overline{s}t \right\}.\ ] ] by ( [ lacapelle - aubareil ] ) we have @xmath270 hence , by cauchy - schwarz inequality , for every @xmath271 we have @xmath272 this gives the equicontinuity of the family ( [ cevennes ] ) . by the way , since @xmath273 , the family is also equibounded . by ascoli theorem we can find a divergent sequence @xmath274 satisfying @xmath275 and and a sequence of minimizers @xmath102 such that the restriction @xmath276}\right.)_{n=1}^{+\infty}$ ] converges uniformly . applying this argument on an increasing and divergent sequence @xmath277 , by a diagonal trick we can find an increasing and divergent sequence of times @xmath274 , a sequence of minimizers @xmath102 and a path @xmath91 such that @xmath278 converges uniformly to @xmath32 on every compact interval . moreover , by lower semi - continuity of the action we have @xmath279}\right.)\le \liminf\limits_{n\rightarrow + \infty } a_l(\gamma_n\left\|_{[0,t]}\right . ) \le at^\frac{1}{3}\ ] ] for every @xmath11 , proving in particular that @xmath280}\right.)$ ] is finite . therefore , @xmath281 is a non - collision configuration for almost all @xmath11 . we prove now that @xmath32 is a minimizing path . since we want to show that @xmath282}\right.$ ] is a minimizer for every @xmath11 , it is sufficient to prove that @xmath282}\right.$ ] is a minimizer for @xmath13 arbitrary great . is obtained by pasting @xmath283 ( reparametrized ) with the straight line joining @xmath281 to @xmath284 we can assume , without loss of generality , that @xmath281 is a non - collision configuration . assuming , for the sake of a contradiction , that @xmath282}\right.$ ] is not a minimizer , there would exists an absolutely continuous path @xmath285\rightarrow { { \mathcal x}}$ ] joining @xmath3 to @xmath281 such that @xmath286}\right.).\ ] ] moreover , there exists @xmath287 and @xmath288 such that @xmath289 where @xmath290 is the closed ball centered in @xmath281 with radius @xmath291 . since the sequence @xmath292}\right.$ ] converges uniformly to @xmath282}\right.$ ] , given @xmath293 there exists a positive integer @xmath294 such that for every @xmath295 we have @xmath296 . let @xmath297\rightarrow { { \mathcal x}}$ ] be the path defined by @xmath298 \\ \frac{t - t}{\epsilon } \gamma(t)+\frac{t - t+\epsilon}{\epsilon } \gamma_{n}(t ) \qquad & \text{if}\quad t\in [ t-\epsilon , t ] , \end{array } \right.\ ] ] where @xmath295 . by construction @xmath299 joins @xmath3 to @xmath300 in time @xmath13 ( see figure [ theorem6 ] ) . moreover , if @xmath301 $ ] , @xmath302 is contained in the ball @xmath303 . computing the action of @xmath299 we get @xmath304 inequalities ( [ la - roque - saint - cristophe ] ) and ( [ la - roque - geac ] ) imply @xmath305}\right.)\ ] ] if @xmath306 is sufficiently small and @xmath307 sufficiently great . this contradicts the minimizing property of @xmath308 and proves that @xmath32 is a minimizer . by marchal theorem , @xmath32 is collision - free ( and in particular it is a real solution of the n - body problem ) for @xmath31 . this complete the proof of theorem [ thm - exist - solut ] . to complete the proof of the main theorem we still have to show that the limit solution @xmath44 is parabolic and asymptotic to @xmath4 . by lemma [ lem - parab - asympt - x0 ] we just need to verify the asymptotic estimates ( [ gamma - i_0-t43 ] ) . we introduce now the following given the functions @xmath309 and @xmath310 , we write @xmath311 as @xmath312 if the quotient @xmath313 is infinitesimal as @xmath312 , uniformly on @xmath314 . in a similar way , we write @xmath315 if the quotient @xmath313 is locally bounded for @xmath316 close to @xmath317 , uniformly on the variables @xmath314 . let us give now a refinement of lemma [ lemme - fondamentale ] . [ nyons ] let @xmath168 and @xmath142 . there exist two constants @xmath318 and @xmath156 such that for every @xmath157 , for every @xmath267 and for every minimizer @xmath143 we have @xmath319 as @xmath320 . proof . * the second inequality is a direct consequence of the first one and of lemma [ nonna - isabella ] . let us prove the first inequality . we consider @xmath171 as a fixed constant , while @xmath13 and @xmath106 are variables . let @xmath155 , @xmath156 and @xmath154 be like in proposition [ lacapelle - marival ] . without loss of generality we can assume @xmath318 . let @xmath157 and @xmath158 . let @xmath321 the path @xmath322\rightarrow { { \mathcal x}},\qquad \overline{\eta}_t(s)=\eta_{t+\tau}\left(\frac{t+\tau}{t}s\right).\ ] ] is a reparametrization of @xmath323 and it joins @xmath170 to @xmath324 in time @xmath13 , thus @xmath325 a computation of the action of @xmath326 gives @xmath327 since @xmath323 is a minimizer joining @xmath170 to @xmath324 in time @xmath328 , by propositions [ lacapelle - marival ] and [ cap - de - l - homy ] we obtain @xmath329 combining inequalities ( [ taybosc ] ) and ( [ cazaubon ] ) , by definition of @xmath323 and @xmath326 we get @xmath330 in a similar way , let us consider a minimizer @xmath331 the path @xmath332\rightarrow { { \mathcal x}},\qquad \overline{\eta}_{t+\tau}(s)=\eta_t\left(\left(\frac{t}{t+\tau}\right)s\right).\ ] ] is a reparametrization of @xmath333 , and it joins @xmath170 to @xmath324 in time @xmath328 , hence @xmath334 arguing as before we get the estimates @xmath335 combining inequalities ( [ barbontan ] ) with ( [ pech - merle ] ) we obtain @xmath336 uniformly on @xmath267 and @xmath337 . with the same argument we find the following estimates @xmath338 replacing ( [ cro - magnon ] ) and ( [ dentelles ] ) into the first inequality of lemma [ lemme - fondamentale ] , since we assume @xmath158 and @xmath156 , we obtain the first inequality of this lemma . this ends the proof . + to simplify the notations we introduce now the functions @xmath339 and @xmath340 [ sommieres ] given @xmath341 and @xmath342 we have @xmath343 with equality if and only if @xmath344 . * proof . * by lemma [ nonna - isabella ] we have @xmath345 with equality if and only if @xmath346 . since @xmath347 is the unique solution of the one - dimensional kepler problem joining @xmath170 to @xmath348 in time @xmath349 ( see lemma [ uniq - solution ] in the appendix ) , it is also a minimizer , therefore @xmath350 with equality if and only if @xmath351 . this proves the lemma . + by homothety invariance , the conclusion of lemmas [ nyons ] and [ sommieres ] can be written in the more compact form @xmath352 as @xmath320 , uniformly on @xmath158 and @xmath337 . + the following theorem is a main tool in the proof of the main theorem . it shows that if @xmath353 is sufficiently small and @xmath349 is sufficiently great , the configuration @xmath22 is close to @xmath354 . [ ardeche ] there exist a function @xmath355\rightarrow { { \mathbb r}}_+$ ] satisfying @xmath356 as @xmath357 , such that for every @xmath358 $ ] , there exists @xmath359 , such that for every @xmath360 , the set of configurations @xmath342 satisfying the inequality @xmath361 is contained in the ball @xmath362 . before giving the proof of theorem [ ardeche ] , we show that this theorem achieve the proof of the main theorem . + * proof of the main theorem. let @xmath83 be the limit solution constructed in theorem [ thm - exist - solut ] and let @xmath102 be the sequence of minimizers uniformly convergent to @xmath32 on every compact interval . let @xmath291 like in theorem [ ardeche ] and @xmath293 . an immediate consequence of inequalities ( [ nesque ] ) is the existence of @xmath363 such that if @xmath364 and @xmath275 we have @xmath365 and by theorem [ ardeche ] @xmath366 for @xmath34 sufficiently great . the sequence @xmath367}\right.$ ] converges uniformly to @xmath282}\right.$ ] as @xmath177 , hence @xmath368 for every @xmath364 . since @xmath369 as @xmath370 , we have proved that @xmath371 that is to say , @xmath32 is parabolic and asymptotic to @xmath4 . this achieves the proof of the main theorem . + the next section is devoted to prove theorem [ ardeche ] . in order to achieve the proof of theorem [ ardeche ] we compare the @xmath0-body problem with a kepler problem on the configuration space with a lagrangian given by @xmath372 let @xmath373 denote the action ( for the lagrangian @xmath374 ) of an absolutely continuous path @xmath118 and @xmath375 the infimum of @xmath373 over all absolutely continuous paths @xmath118 joining @xmath376 to @xmath377 in time @xmath349 . we have the inequality @xmath378 with @xmath379 if and only if there exists a minimizing path ( for the lagrangian @xmath64 ) @xmath380\rightarrow { { \mathcal x}}$ ] joining @xmath376 with @xmath377 such that @xmath381 for every @xmath382 $ ] , and @xmath383 if and only if @xmath376 and @xmath377 are on a same half - line starting from the origin . the function @xmath384 verifies the inequality @xmath385 roughly speaking , to achieve the proof of theorem [ ardeche ] , we replace @xmath353 with @xmath386 and we show that if @xmath306 is small and @xmath349 great , the inequality @xmath387 can be satisfied only if @xmath22 is in a small ball centered in @xmath354 . + this goal will be achieved in two steps . in proposition [ gcors(r ) ] we prove that if @xmath349 is sufficiently great , the set of @xmath388 verifying @xmath389 is contained in a small interval centered in @xmath90 . hence , by inequality ( [ pas - de - la - graille ] ) , the set of configuration @xmath22 verifying @xmath390 is contained in a thin hollow sphere with inner and outer radious close to @xmath90 . in proposition [ applic - lambert ] we show that the set of configurations @xmath22 verifying @xmath387 is a small neighborhood of @xmath354 . [ gcors(r ) ] there exist a function @xmath391\rightarrow { { \mathbb r}}_+$ ] satisfying @xmath392 as @xmath357 , such that for every @xmath393 $ ] there exists @xmath394 , such that for every @xmath395 , the set of @xmath388 satisfying the inequality @xmath396 is contained in the interval @xmath397 $ ] . proof . * by proposition [ elenuccia ] of the appendix there exists @xmath398 and @xmath399 such that for every @xmath400 and for every @xmath401 we have @xmath402 . without loss of generality we will assume @xmath403 . by proposition [ s(c - point ) ] of the appendix we have @xmath404 where @xmath405 as @xmath406 , uniformly on @xmath407 , and where @xmath408 . let us introduce now the function @xmath409 by lemma [ uniq - solution ] the solution joining @xmath170 with @xmath316 in time @xmath261 is monotonic if and only if for @xmath410 , where @xmath411 . we remark that @xmath412 . the energy @xmath413 of this solution is negative if and only if @xmath414 , moreover @xmath415 . let us term @xmath416 . the action @xmath417 is given by @xmath418 hence by lemma [ lem - h - croiss ] , the functions @xmath419 and @xmath420 are of class @xmath421 on @xmath422 , moreover we have @xmath423 proving that @xmath424 is in fact of class @xmath425 on @xmath426 . since the function @xmath427 is increasing and @xmath428 , the function @xmath424 is decreasing for @xmath429 and it is increasing for @xmath430 . the absolute minimum of @xmath424 is achieved at @xmath351 , and we have @xmath431 by lemma [ lem - h - croiss ] , a direct computation of the second derivative of @xmath432 at @xmath90 gives @xmath433 hence , since @xmath434 , there exists @xmath435 and @xmath436 such that @xmath437,\qquad g(r)\ge c_1(r-\alpha)^2.\ ] ] without loss of generality we shall assume @xmath438 and @xmath439 . let @xmath440 and let us define the function @xmath441\rightarrow { { \mathbb r}}_+,\qquad \delta_1(\epsilon)=\sqrt{\frac{2\epsilon}{c_1}}.\ ] ] since @xmath424 is decreasing for @xmath442 and increasing for @xmath443 , for every @xmath393 $ ] we have @xmath444 we come back now to the function @xmath445 . since @xmath446 is infinitesimal for @xmath406 and @xmath447 , for every @xmath393 $ ] there exists @xmath448 such that for every @xmath395 and for every @xmath316 verifying @xmath447 we have @xmath449 . if @xmath395 and @xmath400 we have @xmath450 if @xmath395 and @xmath4510 , \overline{r}[$ ] , by ( [ g - r ] ) , for every @xmath4510,\alpha-\delta_1(\epsilon)[\cup ] \alpha+\delta_1(\epsilon),\overline{r}[$ ] we have @xmath452 this ends the proof of the proposition . + we introduce the following notation : given two configurations @xmath376 and @xmath377 , the angle between @xmath376 and @xmath377 is denoted by the symbol @xmath453 . we always have @xmath454 . [ applic - lambert ] if @xmath455 and @xmath456\rightarrow { { \mathbb r}}_+$ ] are like in proposition [ gcors(r ) ] , there exist @xmath457 $ ] and @xmath458 such that given the function @xmath459\rightarrow { { \mathbb r}}_+ , \qquad \delta_2(\epsilon)=(c_2\epsilon)^\frac{1}{2},\ ] ] for every @xmath358 $ ] , there exists @xmath460 such that for every @xmath461 and for every configuration @xmath342 satisfying @xmath462 we have @xmath463 * proof . * the basic tool of this proof is lambert s theorem . our reference is [ al ] . + let @xmath458 and @xmath457 $ ] . let @xmath464\rightarrow { { \mathbb r}}_+$ ] be the function defined in ( [ delta_2 ] ) . in the following we will ask more precise conditions on @xmath179 and @xmath291 . let @xmath465 , let @xmath22 be a configuration verifying ( [ angle - config ] ) and @xmath341 . the minimizer ( for @xmath374 ) @xmath466\rightarrow { { \mathcal x}}$ ] joining @xmath22 to @xmath467 in time @xmath468 is a collision - free keplerian arc , hence it is contained in the plane generated by @xmath170 , @xmath22 and @xmath467 . introducing a system of polar coordinates in this plane , we can identify @xmath22 with @xmath469 and @xmath467 with @xmath470 where @xmath471 moreover , the path @xmath283 can be written in polar coordinates by @xmath472,\ ] ] where @xmath473 since @xmath283 is collision - free , @xmath474 for all @xmath475 $ ] . by definition of @xmath476 and using the properties of @xmath477 we have @xmath478 we prove now that @xmath283 is a _ direct path _ , that is to say , the total variation of the polar angle @xmath479 is less or equal to @xmath480 . assume , for the sake of contradiction , that @xmath481 . eventually changing the orientation of the plane , we can assume without loss of generality @xmath482 , hence there exists a unique integer @xmath483 and a unique real number @xmath484 $ ] such that @xmath485 the path @xmath486 defined by @xmath487 has the same ends as the original one , moreover @xmath488\int_0^{s-1 } ( \rho^2\dot{\phi}^2)(u)du<0,\ ] ] and we get a contradiction . lambert s theorem state that if @xmath376 and @xmath377 are two configurations and @xmath168 , the action @xmath489 of the direct keplerian arc joining @xmath376 to @xmath377 in time @xmath171 is a function of three parameters only : the time @xmath171 , the distance @xmath490 between the two ends and the sum of the distances between the ends and the origin ( i.e. @xmath491 ) . comparing now @xmath283 with a direct collinear arc , by lambert s theorem we find @xmath492 where @xmath493 moreover @xmath494 where @xmath495 uniformly on @xmath496 and @xmath497 . therefore we get @xmath498 since @xmath499 , applying proposition [ s(c - point ) ] of the appendix to @xmath500 we find @xmath501+g(r,\theta , s),\ ] ] where @xmath502 is infinitesimal as @xmath406 , uniformly on @xmath316 and @xmath503 . in proposition [ gcors(r ) ] we showed that @xmath504 for all @xmath505 . let @xmath506 such that for every @xmath507 , for every @xmath503 satisfying @xmath508 $ ] and for every @xmath509 $ ] we have @xmath510 since the function @xmath511 is decreasing in @xmath512 $ ] , chosing @xmath513 and using the classical expansions of @xmath514 and @xmath515 we find @xmath516-\epsilon \\ & = \epsilon \left [ \beta_0 \alpha^\frac{1}{2 } \left ( 1-\frac{\delta_1(\epsilon)}{\alpha}\right)^\frac{1}{2 } \left ( \frac{c_2 - 4}{8}+\mu(\epsilon)\right)-1 \right ] , \end{array}\ ] ] where @xmath517 as @xmath370 . chosing @xmath518 such that @xmath519 , \qquad \|\mu(\epsilon)\|<\frac{1}{8},\quad \text{and}\quad in such a way @xmath520 we find @xmath521 for every @xmath509 $ ] , for every @xmath503 such that @xmath522 $ ] and for every @xmath461 . this proves the proposition . + the proof of theorem [ ardeche ] is essentially the juxtaposition of the two previous propositions . + * proof of theorem [ ardeche]. + we use the same notations of the previous two propositions . given @xmath358 $ ] , let @xmath523 . by proposition [ gcors(r ) ] and [ applic - lambert ] and by inequality ( [ pas - de - la - graille ] ) , if @xmath360 and @xmath22 is a configuration verifying @xmath390 we have @xmath524 let @xmath525 be the function @xmath526\rightarrow { { \mathbb r}}_+,\qquad \delta(\epsilon)=\left[2\alpha\left(\alpha+\delta_1(\epsilon)\right ) ( 1-\cos\delta_2(\epsilon))+\delta_1(\epsilon)^2\right]^\frac{1}{2},\ ] ] an easy computation show that @xmath369 as @xmath370 and the set of configurations verifying ( [ condit - presque - conclus ] ) is contained in the ball @xmath527 . the theorem is proved . the kepler problem on the half - line @xmath528 is defined by the equation @xmath529 where @xmath530 is the gravitational constant . the lagrangian function of the problem and the energy are written @xmath531 a parabolic solution of the kepler problem is nothing but a solution with zero energy . there is a unique increasing parabolic solution , namely @xmath532 where @xmath533 . given @xmath534 , the energy of a solution connecting @xmath111 to @xmath112 is necessarily greater or equal to @xmath535 . moreover , if @xmath536 , for @xmath537 or @xmath538 there is a unique segment of solution of energy @xmath43 joining @xmath111 to @xmath112 , this solution increases from @xmath111 to @xmath112 . if @xmath539 there are exactly two segments of solutions of energy @xmath43 joining @xmath111 to @xmath112 , a monotonic one , that increases from @xmath111 to @xmath112 , and a non - monotonic one , that increases from @xmath111 to @xmath540 and decreases from @xmath540 to @xmath112 . let @xmath541 be the time employed by the solution of energy @xmath535 to connect @xmath111 to @xmath112 . we have the following lemma , whose proof is left to the reader . [ uniq - solution ] given @xmath534 , and @xmath542 , there exists a unique segment of solution joining @xmath111 to @xmath112 in time @xmath349 , moreover , the solution is monotonic if and only if @xmath543 . given @xmath534 and @xmath542 , we denote by @xmath544 the energy of the unique segment of solution joining @xmath111 to @xmath112 in time @xmath349 , and we denote by @xmath545 the lagrangian action of this solution . since the solution joining @xmath111 to @xmath112 in time @xmath349 is unique , @xmath545 is also the minimum of the action of absolutely continuous paths joining @xmath111 to @xmath112 in time @xmath349 . + we shall study the behaviour of the function @xmath546 for fixed @xmath542 . proof.the parabolic solution @xmath552 has zero energy , hence @xmath553 . since we are interested at what happens when @xmath550 , we assume @xmath554 . by lemma [ lem - h - croiss ] the energy @xmath555 is positive and the solution joining @xmath170 to @xmath316 in time @xmath556 is monotonic . the function @xmath557 verifies the identity @xmath558 where @xmath559 is defined by @xmath560 and it verifies the estimates @xmath561 let us prove now that @xmath562 uniformly on @xmath563 $ ] . assuming , for the sake of contradiction , that ( [ h - infini ] ) is false , there would exist two sequence @xmath564 and @xmath565 $ ] such that @xmath566 is bounded . to simplify notations let us denote @xmath567 . by identities ( [ h - c - eps ] ) and ( [ e - asympt ] ) , the sequence @xmath568 is bounded too . this implies that @xmath569 as @xmath177 . since @xmath570 is continuous and strictly increasing , identity ( [ h - c - eps ] ) gives @xmath571 as @xmath177 . applying again ( [ h - c - eps ] ) and the first of ( [ e - asympt ] ) we obtain @xmath572 that gives a contradiction and proves ( [ h - infini ] ) . writing now ( [ h - c - eps ] ) as @xmath573 using the second of ( [ e - asympt ] ) we obtain the following estimates @xmath574 as @xmath550 , uniformly on @xmath551 $ ] . let us consider now the action @xmath575 . let @xmath576 be the solution joining @xmath170 with @xmath316 in time @xmath556 . we have @xmath577 where @xmath578 is defined by @xmath579 the function @xmath580 verifies the asymptotic estimates @xmath581 replacing ( [ h - asympt ] ) in ( [ s(0,r;1+eps ) ] ) we find ( [ ardechoise ] ) . proof.we first prove that the ( unique ) solution joining @xmath316 to @xmath589 in time @xmath590 is monotonic . in order to simplify the exposition let us term @xmath591 . we shall compare @xmath590 with the time employed by the solution of energy @xmath592 to connect @xmath316 to @xmath593 . as usual we denote @xmath594 this time . by definition of @xmath90 we have @xmath595 where we define @xmath596 since we assume @xmath597 we have @xmath598 an easy computation shows that @xmath599 hence we get the estimates @xmath600 since @xmath601 , we have @xmath602 for @xmath349 sufficiently great , and by lemma [ uniq - solution ] the solution joining @xmath316 to @xmath593 in time @xmath590 is monotonic . + let @xmath603 be the energy of the solution joining @xmath316 to @xmath593 in time @xmath590 . we prove that @xmath604 for @xmath406 , uniformly on @xmath605 , @xmath587 and @xmath588 . the energy @xmath43 satisfies the identity @xmath606 introducing the functions @xmath607 and using the definition of @xmath90 , the relation ( [ rel - h - s - eta - xi ] ) becomes @xmath608 where @xmath609 is defined by @xmath610 we think now at @xmath611 as independent variables . using the implicit function theorem we show that the equation @xmath612 defines a unique @xmath613 function @xmath614 for @xmath615 close to @xmath616 . we observe that @xmath609 is of class @xmath613 with respect to the variables @xmath617 and @xmath618 . moreover @xmath580 is derivable with respect to @xmath22 and @xmath619 @xmath620 is derivable with respect to @xmath22 and @xmath621 , and we have @xmath622 showing that @xmath620 is of class @xmath547 in a neighborhood of @xmath623 . in particular @xmath624 by the theorem of differentiation under the integral sign , @xmath625 , @xmath626 and @xmath627 are well defined , moreover @xmath628 by the way , we have also @xmath629 these computations show that @xmath580 is of class @xmath613 in a neighborhood of @xmath623 . moreover @xmath630 by the implicit function theorem , equation ( [ f(x , y , z , k ) ] ) defines a @xmath613 function @xmath631 in a neighborhood of @xmath616 such that @xmath632 and @xmath633 that is to say @xmath634 coming back to original variables , identity ( [ k - x - y - z ] ) gives @xmath635 as @xmath406 , uniformly on @xmath447 , @xmath587 , and @xmath588 . we compute now the action @xmath636 . since the solution joining @xmath316 to @xmath593 in time @xmath590 ( denoted here @xmath576 ) is monotonic , we have @xmath637 introducing the integration variable @xmath638 , by ( [ h - asympt - c - s ] ) we find @xmath639 where @xmath640 and @xmath641 are the functions defined like in ( [ rocamadour ] ) and @xmath642 where @xmath643 once again , we think at @xmath22 and @xmath621 as independent variables and we give an asymptotic expansion of @xmath644 for @xmath22 and @xmath621 close to @xmath170 . by the classical theorem of differentiation under the integral sign , @xmath645 is derivable in @xmath170 and @xmath646 moreover we have the following estimates for @xmath647 @xmath648 hence @xmath649 as @xmath650 and @xmath651 . replacing in ( [ sarlat ] ) and using ( [ h - asympt - c - s ] ) we find the final estimates ( [ barney - panofsky ] ) . + the two previous propositions imply the following one . proof.*if @xmath654 and @xmath447 , from propositions ( [ s(0,c ) ] ) and ( [ s(c - point ) ] ) we have : @xmath655 therefore @xmath656 uniformly on @xmath551 $ ] . let us consider now the case @xmath657 . forgetting the term @xmath658 in @xmath659 and applying again propositions ( [ s(0,c ) ] ) and ( [ s(c - point ) ] ) we find @xmath660 this estimates implies the limit @xmath661 uniformly on @xmath551 $ ] . + the two limits ( [ lim - c < s13 ] ) and ( [ lim - c > s13 ] ) achieve a proof of the proposition . [ ch1 ] a. chenciner _ collision totales , mouvements compltement paraboliques et rduction des homothties dans le problme des n corps , regular and chaotic dynamics , vol . 3 , @xmath6623 , pp . 93 - 105 , ( 1998 ) _	we consider the @xmath0-body problem in @xmath1 with the newtonian potential @xmath2 . we prove that for every initial configuration @xmath3 and for every minimizing normalized central configuration @xmath4 , there exists a collision - free parabolic solution starting from @xmath3 and asymptotic to @xmath4 . this solution is a minimizer in every time interval . the proof exploits the variational structure of the problem , and it consists in finding a convergent subsequence in a family of minimizing trajectories . the hardest part is to show that this solution is parabolic and asymptotic to @xmath4 .
recently , a lot of attention has been directed to the subject of excitons in bulk crystals due to experimental observation of the so - called yellow exciton series in cu@xmath0o up to a large principal quantum number of @xmath2 @xcite-@xcite . such excitons in copper oxide , in analogy to atomic physics , have been named rydberg excitons . by virtue of their special properties rydberg excitons are of fascination in solid state and optical physics . these objects , whose size scales as the square of the rydberg principal quantum number @xmath3 , are ideally suited for fundamental quantum interrogations , as well as detailed classical analysis . one could expect that rydberg excitons would have been described , in analogy to rydberg atoms , by rydberg series of hydrogen atoms , but it turned out that this generic method of description should has been revised because diameter of such exciton is much larger then wavelength of light needed to create it @xcite . the observation and detailed description of rydberg excitons have opened a new field in condensed matter spectroscopy . in analogy to medium of rydberg atoms , where it has been possible to obtain a large optical nonlinearity at the single photon level and to realize a lot of quantum optics sophisticated experiments such as optical kerr effect or correlated states @xcite , it is expected that the medium of rydberg excitons is also fertile area . the unique combination of their huge size , long radiative lifetimes , possible strong dipole - dipole interaction and , what is the most important advantage for future technological applications , miniaturization of media / samples they are realized in , can be exploited to perform robust light - exciton quantum interfaces for quantum information processing purposes . solid bulk media are systems well worth considering for storing quantum information because they have a number advantages over gases , where a lot of experiments have been done [ for recent review see ref.@xcite ] : they are easy to prepare , diffusion processes are not so fast , much higher densities of interacting particles can be achieved @xcite . a common class of solids used within a quantum information context are rare - earth - metal - doped crystals , where a long time over one minute of storing information has been achieved @xcite , @xcite , and nitrogen - vacancy centres in diamond @xcite which have relatively long spin coherence . however , the size of solid samples used in such experiments is of order of several millimetres . rydberg excitonic samples are much smaller : observation of dipolar blockade in bulk cu@xmath0o @xcite and quantum coherence @xcite were performed in samples as small as several tens of micrometers . realization of these experiments have unlocked the plethora of dynamical effect which might be observed in rydberg excitons media ; one of examples is the electromagnetically induced transparency , the performing of which in cu@xmath0o bulk crystal will be the next step toward potential implementation this medium for quantum information processing . electromagnetically induced transparency ( eit ) @xcite , is one of the important effect in quantum optics as it allows for the coherent control of materials optical properties . the generic eit bases on extraordinary dispersive properties of an atomic medium with three active states in the @xmath4 configuration . this phenomenon leads to the significant reduction of absorption of a resonant probe , weak laser field by irradiating the medium with a strong control field making an otherwise opaque medium transparent . it leads to dramatic changes of dispersion properties of the system : absorption forms a dip called the transparency window and approaches zero while dispersion in the vicinity of this region becomes normal with a slope , which increases for a decreasing control field . the resonant probe beam is now transmitted almost without losses . eit has been explained by destructive quantum interference between different excitation pathways of the excited state or alternatively in terms of a dark superposition of states . since at least 20 years there has been a considered level of activity devoted to eit @xcite , which has been motivated by recognition of a number of its applications among which slowing and storing the light , see i.e. , @xcite-@xcite and references therein , which allows for realization of delay lines and buffers in optical circuits are well - known examples . a remarkable quenching of absorption due to eit in an undoped bulk of cu@xmath0o in a @xmath4-type configuration involving lower levels was examining in @xcite . demonstration of eit in rydberg atoms involving the ladder levels scheme by mohapatra _ et al _ @xcite has taken advantage of their unique properties . in rydberg systems the ladder configuration enables to couple long - living metastable , initially empty upper level with the levels coupled by the probe field . rydberg excitons in cu@xmath0o offer a great variety of accessible states for creating the ladder configuration what enables such a choice of coupling which could be realized by accessible lasers or eventually to be suitable for desirable coherent interaction implementations . rydberg eit has also attracted attention with demonstration of interaction - enhanced absorption imaging of rydberg excitations @xcite . the non - linear response of rydberg medium is proportional to the group index and to the strength of the dipole - dipole interaction and both of them can be extremely large . eit provides a possibility of dissipation - free sensing and probing of highly excited rydberg states and coherent coupling of rydberg states via eit could be used for cross - phase modulation and photon entanglement . rydberg excitons offer an unprecedented potential to study above mentioned phenomena in solids . it seems that , similar to rydberg atoms , strong dipolar interaction between rydberg excitons could appear and leads to so - called photon - blockade , which offers promising means to tailor deterministic single - photon sources @xcite , and to realize photonic phase gates @xcite . up till now the researches involving in rydberg excitons , both theoretical and experimental , have concentrated on their static properties ( excitonic states , electro- and magneto - optical properties ) . the first step toward study the photon blockade , quantum non - linear optics , and many - body physics with rydberg excitons is the realization of rydberg eit . here we focuse our interest on dynamical aspects and properties of rydberg excitons in cu@xmath0o and show , that the rydberg excitonic states can be used to perform the eit . we indicate excitonic states which guarantee of the most efficient realization of the experiment taking into account our previous results concerning excitonic resonaces as well as damping parameters and the matrix elements of momentum operator . our paper is organized as follows . in sec . ii we present the assumptions of the considered model and solve the time evolution equations , obtaining an analytical expression for the susceptibility and the group index . we use the obtained expression to compute the real and imaginary part of the susceptibility and the group index ( sec . iii ) , for a cu@xmath0o crystal slab . we examine in details the changes of both real and imaginary part caused by changes of driving parameters ( for example , the rabi frequencies ) . in sec . iv we draw conclusions of the model studied in this paper and indicate the optimal choice of rydberg states to realize the eit and light slowing . the possible practical importance is also indicated . the phenomenon of eit , described qualitatively above , has been studied theoretically for various configurations of the transitions and probe and control beams . below we propose a theoretical description for the case when the atomic transitions are replaced by intra excitonic transitions in rydberg excitons media . condensed matter exhibits quite a variety of three - level systems where induced transparency could be achieved in much the same way as done with atoms . yet dephasings , which can easily break the coherence of the population trapping state , are typically much faster in solids than in atomic vapors ; it has caused difficulty to observe a large electromagnetic induced transparency effect in solids . this difficulty can be overcame by using the rydberg exciton states . higher states have much larger lifetimes and the dephasing can reach the value which enables observation of the eit effect . below we will consider a cu@xmath0o crystal as a medium where the eit phenomenon can be realized . we use the ladder configuration ( fig . [ fig_schemat ] ) , consisting of three levels @xmath5 , and @xmath6 . as in previous works on rydberg excitons , we focus our attention on the so - called yellow series associated with the lowest inter - band transition between the @xmath7 valence band and the @xmath8 conduction band . because both band - edge states are of even parity , the lowest @xmath9 exciton state is dipole - forbidden , whereas all the _ p _ states are dipole - allowed ; the @xmath9 to @xmath10 transition is also allowed . we have chosen the valence band as the @xmath11 state . the @xmath12 and @xmath13 excitonic states are the @xmath14 and @xmath6 states , respectively . to obtain the @xmath15 state we assume that a constant electric field @xmath16 is applied to the considered system . as it was recently shown , the applied field splits the @xmath17 levels into @xmath15 , @xmath17 , and @xmath18 states@xcite . let the probe / signal field of frequency @xmath19 and amplitude @xmath20 couples the ground state @xmath11 of energy @xmath21 with an excited state @xmath14 of energy @xmath22 . the control field of frequency @xmath23 and amplitude @xmath24 couples a state @xmath6 of energy @xmath25 with the state @xmath14 , as it is illustrated in fig . [ fig_schemat ] . the hamiltonian of such three level system interacting with an electromagnetic wave , in the rotating wave approximation reads @xmath26\vert b\vert\right.\\ & & \left.-\hbar\omega_{2}(z , t)\exp[-{\rm i}(\omega_{2}t - k_{2}z)]\vert a\rangle\langle c\vert+h.c.\right\},\nonumber\end{aligned}\ ] ] + where @xmath27 , @xmath28 , @xmath29 , @xmath30 are the wave vectors and rabi frequencies corresponding to the particular couplings and @xmath31 being the dipole transition moments related to the specific transitions . the time evolution of the system is governed by the von neumann equation with a phenomenological relaxation contribution @xmath32+r\rho,\ ] ] where @xmath33 denotes the density matrix for an exciton at the position @xmath34 , and @xmath35 is the relaxation operator accounting for all relaxation processes in the medium . setting @xmath36,\nonumber\\ \rho_{ac}&=&\sigma_{ac}\exp[-{\rm i}(\omega_{2}t - k_2 z)],\nonumber\\ \rho_{bc}&=&\sigma_{bc}\exp[-{\rm i}(\omega_1-\omega_2)t-(k_1-k_2)z)],\nonumber\\ \rho_{aa}&=&\sigma_{aa},\nonumber\\ \rho_{bb}&=&\sigma_{bb},\nonumber\\ \rho_{cc}&=&\sigma_{cc},\nonumber\end{aligned}\ ] ] one can get rid of time - dependent factors except for slowly varying factors accompanying probe field . we denote by @xmath37 and @xmath38 , respectively the probe and the control beam detunings . while propagation effects for the control field are neglected the evolution of the system is described by the following set of bloch equations @xmath39 parameters @xmath40 , @xmath41 , describe damping of exciton states and are determined by exciton - damping mechanisms comprising , temperature - dependent homogeneous broadening due to phonons and broadening due to structural imperfections and eventual impurities . the relaxation damping rates for the coherence are denoted by @xmath42 , @xmath41 @xcite . it should be noted that in the above equations only the relaxations inside the three level system are considered , so the total probability for the populations of the three levels is conserved : @xmath43 . in order to study propagation of the signal field inside the medium the bloch equations are accompanied by the maxwell propagation equation for the rabi frequency @xmath44 of the probe pulse , which reads in the slowly varying envelope approximation @xmath45 where @xmath46 @xmath47 being the density of excitons , @xmath48 is the transition dipole matrix element , @xmath19 is the electromagnetic wave frequency . in the first order perturbation with respect to the probe field , the evolution of our system reduces to the set of the following equations for the density matrix @xmath49 taking into account that for a weak probe pulse polarization of the medium for a given frequency is proportional to the signal field @xmath50 and to the susceptibility @xmath51 it has a form @xmath52 the steady state complex exciton susceptibility exhibited to the probe field has the form @xmath53 where @xmath54 is the vacuum dielectric constant . the susceptibility is a complex , rapidly varying function of @xmath55 , its real part describe the dispersion and imaginary show the absorption of the medium . due to the dependence of the refractive index @xmath56 on electric susceptibility we define the group index @xmath57 which account on time delay of a pulse propagating in a medium with the group velocity @xmath58 so the slope of dispersion inside the transparency window determines the propagation of the pulse inside the medium . note that derivative of the real part of the susceptibility may be positive ( normal dispersion ) or negative ( anomalous dispersion ) ; in the latter case the group velocity may even become negative . + another way of explaining electromagnetically induced transparency applies the notion of dressed states . consider the subspace spanned by the states a and c ( the energy @xmath25 of the latter being moved by the photon energy @xmath59 ) , coupled by the interaction @xmath60 . the dressed states are eigenvectors of the hamiltonian restricted to this subspace . the eigenenergies are shifted from their bare values ; if the control field is at resonance the shift is equal to @xmath61 . if the probe photon is tuned right in the middle between the dressed eigenenergies , the transition amplitudes from the state b interfere destructively . + in the case of spectrally not too wide pulses one can approximate the susceptibility to the lowest term of its taylor expansion at the line center and in such a case the probe pulse can has the form @xmath62 this means that the pulse moves with the velocity @xmath63 with its shape essentially unchanged apart from an exponential modification of its height and an overall phase shift . the group velocity is approximately the velocity of the pulse maximum ( exactly if there is no damping ) . we have performed numerical calculations for a cu@xmath0o crystal slab with a thickness 30 @xmath64 . a constant electric field @xmath16 is applied in the @xmath34-direction . using the formulas ( [ chiladder ] ) and ( [ group_index ] ) we have calculated the real and imaginary part of the susceptibility and the group index . the values of certain energies , dipole moments and damping parameters characteristic for the excitonic states of fig . [ fig_schemat ] the results from our previous paper @xcite have been used . the detailed calculations of @xmath65 are presented in the appendix . for @xmath47 being the density of excitons we have used the value @xmath66 . applying external electric field @xmath67 one obtains the following values of parameters + @xmath68=3266.576 thz ( = 2150.3 mev ) + @xmath69 thz ( = 20.6714 mev ) + @xmath70 ghz ( = 30 @xmath71ev/@xmath72 ) + @xmath73 ghz ( = 5 @xmath71ev/@xmath72 ) + @xmath74 @xmath75 + @xmath76 as can be seen in fig . [ fig_chi ] a ) for non - zero rabi frequency of the control field the imaginary part of the system s susceptibility reveals a dip in the lorentzian absorption profile called a transparency window . this means that a resonant probe beam which otherwise would be strongly absorbed , is now transmitted almost without losses . the width of the transparency window is proportional to the square of control field amplitude and therefore increasing the control field strength it is possible to open it out . the real part of susceptibility is shown in fig . [ fig_chi ] b ) . the dispersion inside transparency window becomes normal with the slope which increases from a decreasing control field . the normal dispersion inside the window is responsible for reduction of the group velocity the absorption at the resonance does not reaches zero due to finite value of the relaxation rate @xmath77 , but it is very small . this means that the medium has become transparent for a probe pulse which travels with a reduced velocity . it is well known that eit allows us to obtain a steeper slope of the refractive index and thereby a large group index , see fig . [ fig_chi ] c ) . the negative values of @xmath78 , corresponding to the regions of anomalous dispersion and high absorption , have been omitted for clarity . [ fig_chi ] d ) shows how the group index and absorption inside the transparency window depend on the rabi frequency @xmath60 . the plots represent a cross - sections of fig . [ fig_chi ] a ) and fig . [ fig_chi ] c ) at @xmath79 . for @xmath80 , the susceptibility given by eq . ( [ chiladder ] ) has a single resonance , so that the imaginary part of susceptibility is significant . as the rabi frequency @xmath60 increases , a transparency window is formed . for considered excitonic transitions optimal slowing down of order @xmath81 appears for transparency window of width of tens ghz ; as shown on the fig . [ fig_chi ] d ) , there is some optimum @xmath82 where the group index has a maximum value , corresponding to a narrow , but fully formed transparency window with significant dispersion @xmath83 . by increasing the @xmath60 further , one obtains a wider window , characterized by higher group velocity but also smaller absorption . it should be stress that the frequency of the signal to be slowed down or even processed determines the choice of the exciton level involved @xmath14 . the influence of the damping rate @xmath84 on the group index and on the group velocity is not crucial . its increase causes the widening of transparency window which is accompanied by decrease of the dispersion . a ) group velocity index @xmath85 . b ) real part of susceptibility @xmath86 . c ) imaginary part of susceptibility @xmath87 . d ) group index @xmath88 and @xmath89 , in the center of the transparency window.,title="fig : " ] b ) . a ) group velocity index @xmath85 . b ) real part of susceptibility @xmath86 . c ) imaginary part of susceptibility @xmath87 . d ) group index @xmath88 and @xmath89 , in the center of the transparency window.,title="fig : " ] c ) . a ) group velocity index @xmath85 . b ) real part of susceptibility @xmath86 . c ) imaginary part of susceptibility @xmath87 . d ) group index @xmath88 and @xmath89 , in the center of the transparency window.,title="fig : " ] d ) . a ) group velocity index @xmath85 . b ) real part of susceptibility @xmath86 . c ) imaginary part of susceptibility @xmath87 . d ) group index @xmath88 and @xmath89 , in the center of the transparency window.,title="fig : " ] the universal rydberg nature can be exploited in systems other than atomic gases . the impressive observation of a rydberg blockade shift on a very different platform offers a new approach for studying semiconductor systems and also provides entirely new long - term perspectives for developing novel devices , which are more robust and compact than atomic systems . we have indicated the optimal states and well justified parameters to attempt the observation eit in rydberg excitons cu@xmath0o media which allow to obtain considerable value of group index . due to coherence properties of rydberg excitons the manipulations of the medium transparency is possible ; the width of the window and slowing down the group velocity of the pulse travelled inside the sample might be changed in controlled way by the strength of control field . the ability to control on - demand group index enables storing and retrieving light pulses , which is a basis to quantum memory implementation . the way of a precise dynamical control of the optical properties of the medium by optical means reveals new aspects of excitons quantum optics and is supposed to lead to constructing efficient tools for photonics , e.g. , delay lines , quantum switches or multiplexers . so far experimental demonstration of eit in rydberg excitons media is probably difficult to realize , but it seems safe to expect such experiments in the future . performing eit in excitonic rydberg media will be the step toward realization of controlled interaction of rydberg excitons in integrated and scalable solid state devices . for the 2@xmath17 exciton we take the 2@xmath90 state , i.e. @xmath91 state , which couples with the @xmath34- component of the electromagnetic wave . so all the waves propagating in the considered medium must have the @xmath34- component . according to the notation of ref . @xcite the energy of 2@xmath90 exciton follows from the relations + @xmath92 with @xmath93 where for @xmath94 and in units @xmath95 , one obtains @xmath96 the energy eigenvalues have the form @xmath97 where @xmath98 , with @xmath99 defined by with @xmath100 @xmath101 being the exciton effective masses anisotropy parameter , @xmath102 are the laguerre polynomials , and @xmath103 being the spherical harmonics . the energy for the @xmath13 exciton results from eq . ( [ energy2p ] ) where we take the larger value from the two solutions @xmath104 . + the energy for @xmath12 exciton could be calculated from the equation @xmath105 where we take the smaller value from the solutions . the dipole matrix element @xmath106 follows from the relation @xmath107 @xmath108 being the longitudinal - transversal splitting , and @xmath109 the so - called coherence radius . positions of our resonances were in perfect agreement with the experimental data ( see fig . [ fig_comp ] a ) the damping parameters were determined by fitting our results regarding to the linewidths of resonances . we were able to estimate the dissipation rates of the p excitons ( fig . [ fig_comp ] a ) and s excitons ( fig . [ fig_comp ] b ) which we later used in our numerical calculations . the relaxation rates are comparable to the values found in prior literature @xcite . t. kazimierczuk , d. frhlich , s. scheel , h. stolz , and m. bayer , nature * 514 * , 344 ( 2014 ) . s. hfling and a. kavokin , nature * 514 * , 313 ( 2014 ) . j. thewes , j. hecktter , t. kazimierczuk , m. amann , d. frhlich , m. bayer , m. a. semina , and m. m. glazov , phys . * 115 * , 027402 ( 2015 ) . j. hecktter , _ stark - effect measurements on rydberg excitons in cu@xmath0o_. thesis . technical university dortmund , 2015 . s. zieliska - raczyska , g. czajkowski , and d. ziemkiewicz , phys . b * 93 * , 075206 ( 2016 ) . f. schweiner , j. main , and g. wunner , phys . b * 93 * , 085203 ( 2016 ) f. schne , s .- o . krger , p. grnwald , h. stolz , m. amann , j. hecktter , j. thewes , d. frhlich , and m. bayer , phys . b * 93 * , 075203 ( 2016 ) . m. feldmaier , j. main , f. schweiner , h. cartarius , and g. wunner , j. phys . b : at . mol . * 49 * , 144002 ( 2016 ) . f. schweiner , j. main , m. feldmaier , g. wunner , and ch . uihlein , arxiv:1604.00492v1 [ cond-mat.mtrl-sci ] 2 apr 2016 ; phys . b , accepted ( 2016 ) . m. amann , j. thewes , d. frlich , and m. bayer , nature materials , * 15 * , 741 ( 2016 ) . p. grnwald , m. amann , j. hecktter , d. frhlich , m. bayer , h. stolz , and s. scheel , phys . lett . * 117 * , 133003 ( 2016 ) . s. zieliska - raczyska , d. ziemkiewicz , and g. czajkowski , phys . b * 94 * , 045205 ( 2016 ) . o. firstenberg , c. s. adams , and s. hofferberth , j. phys , b : at . mol , opt . phys . * 49 * , 152003 ( 2016 ) . m. johnsson and k. molmer , phys . a * 70 * , 032320 ( 2004 ) . g. heinze , ch . hubrich , and t. halfmann , phys . . lett . * 111 * , 033601 ( 2013 ) . g. d. fuchs , g. burkard , p. v. klimov , and d. d. awschalom , nature physics * 7 * , 789 - 793 ( 2011 ) . d. schraft , m. hain , n. lorenz , and t. halfmann , phys . lett . * 116 * , 073602 , ( 2016 ) . s. e. harris , phys . today * 50 * , 36 ( 1997 ) . m. fleischhauer , a. immamoglu , and j. p. marangos , rev . 77 * , 633 ( 2005 ) . a. raczyski , j. zaremba , and s. zieliska - kaniasty , phys . a * 69 * , 043801 - 5 ( 2004 ) . m. perdian , a. raczyski , j. zaremba , and s. zieliska - kaniasty , optics comm . * 266 * , 552 - 557 ( 2006 ) . a. raczyski , j. zaremba , and s. zieliska - kaniasty , phys . a * 74 * , 031810 - 7 ( 2007 ) . m. artoni , g. c. la rocca , and f. bassani , europhys . lett . * 49 * , 445 , 2000 . a. k. mohapatra , t. r. jackson , and c. s. adams , phys . lett . * 98 * , 113030 ( 2007 ) . g. gnter , m. robert - de - saint - vincent , h. schempp , c. s. hofmann , s. whitlock , and m. weidemller , phys . lett . * 108 * , 013002 ( 2012 ) . t. peyronel , o. firstenberg , qi - yu liang , s. hofferberth , a. v. gorshkov , t. pohl , m. d. lukin , and v. vuleti , nature * 488 * , 57 ( 2012 ) . d. paredes - barato and c. s. adams , phys . lett . * 112 * , 040501 ( 2014 ) . j. hecktter , private correspondence .	we show that the electromagnetically induced transparency ( eit ) is possible in a medium exhibiting rydberg excitons and indicate the realistic parameters to perform the experiment . the calculations for a cu@xmath0o crystal are given which show that in this medium due to large group index one could expect slowing down a light pulse by a factor about @xmath1 .
coagulation of dust particles by mutual collisions is a fundamental process occuring in protoplanetary disks . it marks the first steps of planet formation . the formation of millimeter size aggregates from micron - sized dust particles proceeds rapidly . as collision velocities in this range are small and particles stick together easily , the growth of aggregates is indisputable @xcite . with increasing particle size , also the relative velocities between particles increase and velocities reach values up to 60 m / s for meter size bodies or even more , eventually @xcite . with increasing velocities collision results become more complex and compaction , fragmentation and rebound are typical outcomes @xcite . how the formation of larger objects proceeds from here is not settled yet . one way might be a continuous process of sticking collisions . experimental studies show that under certain conditions growth is still possible even in the high velocity range with velocities of up to 60 m / s @xcite . in fact fragmentation assists growth at high speed impacts , as a large amount of the impact energy is dissipated by fragmentation . the experiments were somewhat arbitrarily based on collisions into compact targets with volume filling factors of 33% . the volume filling factor is defined as the ratio between the volume filled by dust to the total volume of a dust aggregate or @xmath6 where @xmath7 is the density of the aggregate and @xmath8 is the density of the used material . however , the mechanical properties of the target bodies determine the outcome of a collision and the mechanical properties of dust agglomerates are tied to the volume filling factor of the aggregates @xcite . beyond direct sticking , several authors propose different instability and gravity driven trapping and concentration scenarios to form planetesimals or even larger bodies @xcite . these mechanisms strongly depend on the gas - grain coupling times of large aggregates . the gas - grain coupling time depends on the mass over surface ratio . that means that the particle size for which these mechanisms work depends on the porosity , so porosity is also important in this context . the basic question behind this work therefore is how porosities evolve through the collisional history of growing aggregates . initially dust growth proceeds as fractal growth @xcite . with increasing particle size the collision velocities increase and particles become compacted during collisions @xcite . according to @xcite , decimeter bodies gain most of their mass by accretion of small ( 50 to 100 @xmath9 m ) particles at collision velocities of about . therefore , the basic process which we simulate in laboratory experiments , reported here , is that a body grows by continuous impacts of aggregates in this size range @xmath10 . in a first experimental study of this kind @xcite investigated the structure of evolving decimeter size bodies , which form by multiple impacts of small particles at collision velocities of 7.7 m / s . they showed that dust agglomerates forming by direct sticking and reaccretion have a constant volume filling factor of @xmath11 . the collision history is erased during this growth . independent of the special morphology of the surface after an impact , the top layer is compacted to the same level by following impacts for targets of mm to cm size . the experiments by @xcite were performed with one collision velocity only to determine the properties of decimeter aggregates . smaller bodies are expected to form at lower impact speed though and the porosities might evolve differently . just how the dependency of the volume filling factor on collision velocity looks like has been unclear . @xcite found a rather strong increase of the volume filling factor with collision velocity for multiple impacts . however , their projectiles were larger ( mm ) , highly porous , did always hit the same spot and consisted of spherical particles . this situation is different for aggregates building a surface in a random way through collisions as in @xcite . within this work we present experiments in analogy to our earlier study in @xcite forming bodies by multiple impacts of small particles but for varying velocities , which we regard as a possible scenario how macroscopic bodies evolve by accretion of small particles at typical collision speeds between 1.5 m / s and the experiments are carried out with a similar method as the experiments already described by @xcite . small compact dust agglomerates ( @xmath12 volume filling factor ) collide with solid targets . the experimental setup is shown in fig . [ setup ] . the central part of the experiment is a sieve with a mesh width of and a rotating stirrer inside . due to the stirrer a constant beam of dust agglomerates leaves the sieve and is accelerated by gravity . the experiments are carried out at pressures below to neglect the influence of gas drag . the drop height is adjustable between and , which leads to impact velocities between and . the collisions are observed with a video camera ( 25 frames / s ) and illuminated by a stroboscope at a flash frequency of each frame therefore is a multiple exposure , so the particle trajectories are directly visible . from the distance between two particle images the impact velocities can be derived directly with the known time interval between two flashes of the stroboscope . due to the launch mechanism the projectiles have vertical trajectories . the impact ejecta leave the target under varying angles . the horizontal @xmath13 and the vertical @xmath14 component of the ejecta velocities are determined by fitting parabolas to the trajectories . due to the vacuum only gravity ( @xmath15 ) acts on the ejected particles . after the first impacts a dust cover forms on any arbitrary substrate used as target . all following impacts are pure dust on dust collisions . targets grow by a combination of direct sticking and reaccretion of ejecta due to gravity . the latter can be regarded as analog for gas driven reaccretion in protoplanetary disks @xcite . there , collisions between large dust aggregates ( few cm or larger ) with small ( @xmath16 m ) particles are induced by their different coupling to the surrounding gas . impact ejecta therefore feel the relative motion of the larger body with respect to the gas as head wind , which drives them back to the target surface . @xcite showed that gravity in the laboratory simulates the acceleration of impact fragments by the gas drag very well , as it is of the same order . while details of accretion efficiencies depend on the target size and gas density the effect will also be present in protoplanetary disks . in the experiments we used metal cylinders of eight different sizes as targets . therefore , ejecta of a given velocity might be reaccreted on the large targets but not on the smallest one . this way , we get a variation in the ratio between direct sticking and reaccretion of ejecta . the target sizes range from 3.2 mm to 36.6 mm in diameter . the growing targets are very similar in their shape . a collection of targets of different sizes and grown at different impact velocities is shown in fig . [ targets ] . the targets can be described as cones with their slopes decreasing to the top . to determine the volume from the images , the projected surface is analyzed . it is assumed that each target consists of circular disks stacked upon each other , which might be shifted with respect to each other . with this assumption and with @xmath17 circular disks with a diameter @xmath18 and a thickness @xmath19 , the volume of the target can be calculated via @xmath20 for the analysis of the images , the thickness of the disks is set to one pixel . for each target several images from different sides are analyzed and this method is repeated for each image and for all eight different targets . all experiments described in this work are carried out with irregularly shaped quartz dust . the sizes of single monomers range from to with of the mass being between and . the same dust material has already been used in numerous experimental studies on protoplanetary growth @xcite . the particles are launched with a sieve with a mesh width of , which is the same mesh width used by @xcite . due to the dimensions of the drop tube , the drop height can be adjusted between 0.1 m and 2.5 m , which leads to impact velocities between 1.5 m / s and 7 m / s . the projectiles have varying initial velocities due to the launching process . within this study four different drop heights ( 0.1 m , 0.5 m , 1.5 m , and 2.5 m ) and therefore four impact velocities have been analyzed . [ velocities ] shows the velocity distributions measured for the different drop heights . the mean velocities and their standard deviations have been determined as , , , and for increasing drop heights , respectively . all collisions lead to a combination of direct sticking and reaccretion of rebounded or ejected aggregates due to gravity . @xcite showed that the size distribution of the ejected dust agglomerates does not vary significantly from the initial size distribution though fragmentation was observed upon impact . a large amount of smaller fragments ( few @xmath1 m ) form , but they contribute only 5 % to the total fragment mass and therefore are of minor importance . as we study slower collisions here , we assume that this still holds . the velocity distributions for impact ejecta measured for the different impact velocities are shown in fig . [ ejecta ] . the impact fragments are ejected from the impact site in arbitrary directions . as the ejecta velocities are derived from the images of only one camera only one component of the horizontal velocity of ejecta is measured . the complete horizontal velocity of the impact ejecta consists of two horizontal components and is therefore expected to be larger than the measured values by a factor of @xmath21 , assuming that both horizontal components are the same . as most mass of the ejecta is in particles comparable in size to the impacting aggregates the collisions might be characterized by a coefficient of restitution , which we define as @xmath22 . for this both horizontal components of the ejecta velociy are assumed to be the same on average due to symmetry . the velocities of the ejecta hardly change with increasing impact velocities , as already shown in fig . [ ejecta ] . this means that the coefficient of restitution decreases with impact velocity . the coefficient of restitution and its dependency on the collision speed is shown in here , the mean impact and ejecta velocities are used to calculate the coefficient of restitution and the errorbars are determined using the standard deviations for the velocities . the grey marked data point at is the result of the experiments by @xcite . with increasing impact velocities obviously more kinetic energy is dissipated , so the velocities of ejecta stay small . there are two processes which might be responsible for the increasing energy loss . one might be the formation of a large number of very small fragments as found in the size distribution by @xcite . however , these small fragments play only a minor role for the total mass of the particle . the other effect is internal restructuring of the impacted target , which should lead to a lower porosity of the evolving aggregates with increasing collision velocity . the porosities have been determined for targets of ( eight ) different sizes , using images from different perspectives for the volume determination as already described in section [ methods ] . the mean porosities or volume filling factor ( 1 - porosity ) of the analyzed targets and their standard deviations for each impact velocity are shown in . the grey marked data point is the value determined by @xcite . the volume filling factor of the forming dust aggregates increases with the impact velocity . this is expected , as the dynamic pressure of the colliding projectiles increases with the impact velocity . however , the volume filling factor saturates at impact velocities above 5 m / s and reaches a constant value of about @xmath3 . the plotted curve in is a function similar to the fermi - function , which has already been used to charcterize the compression behaviour of dust agglomerates by @xcite and @xcite . this function of the impact velocity @xmath23 is given by @xmath24 here , @xmath25 is the minimum porosity for dust samples prepared by sieving without any further compression and is set to @xmath26 , which is in agreement with previous studies by @xcite . the upper compression limit is set to @xmath27 according to the experiments described here . this is close to the maximum value achieved by local compression as already proposed by @xcite . the other two parameters are adjusted to fit the curve to the data points . here , @xmath28 describes the velocity for which the mean value between @xmath25 and @xmath29 is reached and is found to be @xmath30 . the other parameter describes the steepness of the transition between the minimum and the maximum volume filling and is found to be @xmath31 . the compression with increasing impact velocity obviously is well represented by the used function . the saturation volume filling factor is much larger than the volume filling factor feasible by uni - directional compression as determined by @xcite . in that case the maximum compression reachable with the dust material used in this work is @xmath32 . this can be understood as in our case the dust bed is much larger than the impacting aggregates and the dust bed surrounding the impact side provides support for higher compression at the impact side . however , at the same time higher filling factors are prohibited as the target gets destroyed in the close vicinity of the impact side if more pressure is applied which would lead to a pressure relief and keep the porosity at a constant level . it is quite clear that higher compression ( higher volume filling factors ) in principle are possible , as e.g. @xcite measured a maximum volume filling of @xmath33 but only if pressure is applied to all sides of a target . the impact experiments presented here suggest that the dynamic pressure induced by incoming small projectiles is neither uni - directional , nor omni - directional , but somewhere between . this is in agreement with the dust agglomerates used as targets and projectiles in the collision studies by @xcite and @xcite . the porosity in the exeriments by @xcite produced volume filling factors up to @xmath34 at 6 m / s . a difference and possibly a higher value for the saturation volume filling factor might result from the difference in monomer choice . spherical and monodisperse particles as used by @xcite can be compressed more easily . this has e.g. been shown by experiments on uni - directional compression by @xcite . the limit in volume filling factor found by @xcite for spherical particles was @xmath35 . within the error bars the largest value by @xcite is still well above our saturation value . the porosity dependence on collision velocity in the data by @xcite is steeper and they extrapolated the data to volume filling factors of @xmath36 at about 10 m / s . our experiments with irregular shaped grains , which would be expected in protoplanetary disks show , that the volume filling factor levels off at much lower values and lower velocities . we therefore regard such highly compressed targets as unlikely if an aggregate grows by addition of small aggregates and find that a value of @xmath3 describes the volume filling factor of realistic arbitrarily shaped dust aggregates in a wide range . in the reported experiments the targets grow by a combination of direct sticking and reaccretion due to gravity . however , gravity in the laboratory replaces the gas drag in protoplanetary disks . large dust aggregates move through the disk and feel a constant head wind corresponding to their drift and sedimentation speeds . impact ejecta are exposed to this head wind and are accelerated back towards the target surface . this acceleration can be treated as constant to the first order and is given by @xmath37 with the wind speed @xmath38 and the coupling time @xmath39 . the wind speed is equivalent to the collision velocitity . according to @xcite , the coupling time is given by @xmath40 here , @xmath41 is the gas density , @xmath42 is the thermal velocity of the gas molecules given as @xmath43 here , @xmath44 is the gas constant , @xmath45 g / mol is the molar mass of the gas ( mixture of hydrogen and helium ) , @xmath46 is the gas temperature , and @xmath47 is the ratio between particle mass and the cross section of the particle . for a spherical particle with a density @xmath48 and a volume filling factor @xmath49 it is given as @xmath50 . to estimate the gas drag on the impact ejecta we use a volume filling of @xmath3 . the quartz dust used has a density of @xmath51 g/@xmath52 . to estimate the acceleration by the head wind we consider the minimum mass solar nebula ( mmsn ) model by @xcite . the gas density is @xmath53 and the temperature is @xmath54 for a particle of 100 @xmath1 m radius with the given material parameters at 1 au radial distance these equations lead to a coupling time of about . for the lowest collision velocity ( head wind velocity ) of 1.5 m / s this leads to an acceleration of , which is smaller than gravity in the laboratory by a factor of 200 . therefore , the influence of reaccretion in a disk is reduced . nevertheless , reaccretion even of large 100 @xmath1 m fragements for cm - size aggregates in protoplanetary disk is still possible . this can be seen from simple estimates of the distance traveled by an ejected / rebounding aggregate before it hits a surface again with the a given acceleration . a particle launched from a surface with a vertical velocity of 0.1 m / s ( see fig . [ ejecta ] ) reaches the surface again after a time of @xmath55 or 4 s. if the particle travels with the same horizontal velocity it hits the surface a second time at a distance of 40 cm from the launch site . in a mmsn collision velocities of 1.5 m / s correspond to a particle size of about 2 cm , according to @xcite and @xcite . therefore , the considered particle will not be reaccreted but misses the target and is transported away . this might not support the idea of reaccretion but we started with the typical values on purpose to show that even for the average particle the average acceleration is _ only _ a factor of 200 to low . we clearly note that the influence of reaccretion depends on the gas density , ejecta size and ejecta velocity . the values above have been determined for the mmsn by @xcite . other disk models , like e.g. developed by @xcite or by @xcite , give slightly different values , as they use different density and temperature profiles for their disk models . for particles at 1 au radial distance to the protostar the gas densities and temperature profiles given by @xcite or by @xcite will lead to different coupling times . in case of the model by @xcite no reaccretion by gas drag will occur , as the gas density in this model is very small and leads to a coupling time of @xmath56 h. on the other hand , the disk model by @xcite gives a much denser disk and leads to a coupling time of @xmath57 s for a dust agglomerate of @xmath58 m in size at 1 au . this is already sufficient for reaccretion , as the traveling distance is already reduced to 1.6 cm , which will already enable small bodies to reaccret material by gas drag . additionally , the radial distance to the central star is also very important for the reaccretion process . in case of a radial distance of 0.5 au , the coupling time of a 100 @xmath1 m dust agglomerate is only 4.2 s in the mmsn by @xcite . this already reduces the traveling time of ejected fragments to 0.36 s , which leads to a travelled distance of 5.6 cm for a fragment of 100 @xmath1 m in size . in case of smaller fragments ( 10 @xmath1 m ) this distance can easily be reduced to values of 5 mm . in this case reaccretion of impact ejecta will work more efficiently . growth has been observed on targets of different sizes with diameters in the range between 3.2 mm and 35 mm and the resulting volume filling is qualitatively the same . with a typical horizontal velocity of 0.1 m / s and a typical size of 100 @xmath1 m , ejecta will travel 2 mm before lading on the target surface again . the horizontal velocities typically are larger than the measured horizontal comonent due to projection effects ( only one camera ) . the assumption of 0.1 m / s is therefore rather low and ejecta will travel further than 2 mm . this leads to the conclusion that the amount of reaccreted particles is significantly smaller for the small ( 3.2 mm ) targets than for the larger ones . similar experiments by @xcite show similar results . their experiments showed that aggregates growing on a thin glass edge ( 1.1 mm thick ) have the same volume filling as agglomerates growing within a beaker or on the outer shell of a rotating cylinder . @xcite showed that there is no qualitative difference between complete reaccretion ( beaker ) , small amount of reaccretion ( glass edge ) or no reaccretion ( rotating cylinder ) . it was shown in previous experiments @xcite that reaccretion of small particles does not only work in the molecular flow regime in which the streamlines are directed to the target surface , but also in the transition regime between the molecular flow regime and the hydrodynamic regime . reaccretion was found to be an efficient mechanism for growth of dust agglomerates . the target shapes give evidence that direct sticking and reaccretion occur for all target sizes , as all grown targets have the same cone shape , independent of the target size or the impact velocity ( fig . [ targets ] ) . @xcite developed a toy model for target growth by reaccretion . they showed that the probability for reaccretion is largest at the center of a target and becomes smaller to the edges . the result is a shape which is found in all our collision experiments of this type , as already shown in the distance travelled by ejecta before hitting a second time can be reduced easily by a factor of 20 by considering either slightly smaller particles ( projectile fragmentation ) , slightly smaller ejecta velocities , or slightly larger gas densities . the first two are present in the size and velocity distributions ( @xcite , fig . [ ejecta ] ) . the gas density is certainly higher and high enough for accreting almost all ejecta slighly inwards of 1 au . so growth of cm aggregates to larger aggregates will occur if smaller aggregates are present . the influence of the target size and the ratio between directly sticking aggregates and reaccreted ejecta at different velocities on the resulting volume filling is analyzed in no significant dependency between the target size and the measured volume filling factor can be determined for any collision velocity , which is in agreement with the results by @xcite . the observed porosities therefore can be regarded as universal being independent of the target size but only depending on the collision velocity . within this work the self consistent growth of macroscopic dust agglomerates by accretion of small dust aggregates has been analyzed . experiments were carried out to bridge the size range between sub - mm particles and the decimeter size range , which is critical for a variety of growth models . in general we conclude that the reacccretion by gas drag and therfore the growth efficiency might vary strongly for different locations in disks , but that the growth of larger aggregates is possible in a growth model where sub - mm particles impact a growing aggregate . together with earlier experiments by @xcite and @xcite we find that growth is possible up to collision velocities of at least 60 m / s . * dust agglomerates of a few centimeters in size or larger grow by a combination of direct sticking and reaccretion , if they are exposed to multiple impacts of sub - mm particles . * reaccretion by gas drag can aid protoplanetary growth efficiently . * large aggregates have a maximum volume filling of about @xmath3 . although the evolving dust agglomerates become more compact when the impact velocities increase , a saturation is reached for impact velocities of more than 5 m / s . * the coefficient of restitution decreases with increasing impact velocities in the studied velocity regime up to about 8 m / s as the ejecta velocity is essentially independent of the impact velocity . the energy is dissipated by compaction of the target surface and by projectile fragmentation . beitz , e. , gttler , c. , blum , j. , meisner , t. , teiser , j. , wurm , g. 2011 , , submitted . blum , j. , wurm , g. , kempf , s. , & henning , t. 1996 , , 124 , 441 blum , j. , & wurm , g. 2000 , , 143 , 138 blum , j. , schrpler , r. , davidsson , b. j. r. , & trigo - rodrguez , j. m. 2006 , , 652 , 1768 blum , j. , & wurm , g. 2008 , , 46 , 21 brauer , f. , dullemond , c. p. , & henning , t. 2008 , , 480 , 859 desch , s. j. 2007 , , 671 , 878 dominik , c. , & tielens , a. g. g. m. 1997 , , 480 , 647 goldreich , p. & ward , w. r. , 1973 . . , 183 , 10511062 . gttler , c. , krause , m. , geretshauser , r. j. , speith , r. , & blum , j. 2009 , , 701 , 130 gttler , c. , blum , j. , zsom , a. , ormel , c. w. , & dullemond , c. p. 2010 , , 513 , a56 hayashi , c. , nakazawa , k. , & nakagawa , y. 1985 , protostars and planets ii , 1100 johansen , a. , & youdin , a. 2007 , , 662 , 627 johansen , a. , oishi , j. s. , mac low , m .- m . , klahr , h. , henning , t. , & youdin , a. 2007 , , 448 , 1022 kothe , s. , gttler , c. , & blum , j. 2010 , , 725 , 1242 paraskov , g. b. , wurm , g. , & krauss , o. 2007 , , 191 , 779 safronov , v. s. , 1969 . schrpler , r. , & henning , t. 2004 , , 614 , 960 sekiya , m. , & takeda , h. 2003 , earth , planets , and space , 55 , 263 suyama , t. , wada , k. , & tanaka , h. 2008 , , 684 , 1310 teiser , j. , & wurm , g. 2009 , , 393 , 1584 teiser , j. , & wurm , g. 2009 , , 505 , 351 wada , k. , tanaka , h. , suyama , t. , kimura , h. , & yamamoto , t. 2008 , , 677 , 1296 weidenschilling , s. j. 1997 , , 127 , 290 weidenschilling , s. j. , & cuzzi , j. n. 1993 , protostars and planets iii , 1031 weidling , r. , gttler , c. , blum , j. , & brauer , f. 2009 , , 696 , 2036 wurm , g. , paraskov , g. , & krauss , o. 2005 , , 71 , 021304 wurm , g. , paraskov , g. , & krauss , o. 2005 , , 178 , 253 wurm , g. , blum , j. , & colwell , j. e. 2001 , , 151 , 318 wurm , g. , blum , j. , & colwell , j. e. 2001 , , 64 , 046301 youdin , a. , & johansen , a. 2007 , , 662 , 613 zsom , a. , ormel , c. w. , gttler , c. , blum , j. , & dullemond , c. p. 2010 , , 513 , a57	aggregation of dust through sticking collisions is the first step of planet formation . basic physical properties of the evolving dust aggregates strongly depend on the porosity of the aggregates , e.g. mechanical strength , thermal conductivity , gas - grain coupling time . also the outcome of further collisions depends on the porosity of the colliding aggregates . in laboratory experiments we study the growth of large aggregates of @xmath0 3 mm to 3 cm through continuous impacts of small dust agglomerates of 100 @xmath1 m size , consisting of @xmath1 m grains at different impact velocities . the experiments show that agglomerates grow by direct sticking as well as gravitational reaccretion . the latter can be regarded as suitable analog to reaccretion of fragments by gas drag in protoplanetary disks . experiments were carried out in the velocity range between 1.5 m / s and 7 m / s . with increasing impact velocities the volume filling factor of the resulting agglomerates increases from @xmath2 for 1.5 m / s to @xmath3 for 7 m / s . these values are independent of the target size . extrapolation of the measured velocity dependence of the volume filling factor implies that higher collision velocities will not lead to more compact aggregates . therefore , @xmath3 marks a degree of compaction suitable to describe structures forming at @xmath4 . at small collision velocities below 1 m / s highly porous structures with @xmath5 will form . for intermediate collision velocities porosities vary . depending on the disk model and resulting relative velocities , objects in protoplanetary disks up to dm - size might evolve from highly porous ( @xmath5 ) to compact ( @xmath3 ) with a more complex intermediate size range of varying porosity .
the magnetization dynamics of single - domain nanometric particles at low temperature is presently a subject of intense interest , in the hope of finding evidence for quantum tunneling of the magnetic moment through the anisotropy barrier associated with the particle @xcite . apart from some pioneering attempts at a study of a unique particle @xcite , most efforts are concentrated on macroscopic samples , in which an accurate knowledge of the effective distribution of barriers is difficult , hence hindering a clear interpretation of the results @xcite,@xcite . moreover , except in a few cases @xcite , the low - temperature range of the published data is often limited to pumped - he cryogenic techniques ( @xmath0 ) , which still makes an unambiguous characterization of quantum effects more difficult . in this paper , we present magnetic measurements which have been performed using a dilution refrigerator @xcite , that allow data to be taken down to @xmath1 . we have studied a sample of particles , dispersed in a silica matrix , with a typical diameter of @xmath2 . the relaxation dynamics of particles has already been shown to exhibit some anomalies @xcite , that appear at the very end of the accessible temperature range ( 1.8 k ) . our present data show that the relaxation rates in our sample do indeed fail to go down to zero when the temperature is lowered to @xmath3 . the small particles of ( maghemite ) are embedded in a silica matrix , obtained by a polymerization process at room temperature . they are diluted to a volume fraction of @xmath4 , in order to have them as independent as possible . the diameter distribution obtained by transmission electron microscopy is shown in the inset of fig . 1 ; it can be fitted to a log - normal shape with peak value @xmath5 and standard deviation @xmath6 . fig.1 presents magnetic characterization measurements performed with a commercial squid magnetometer ( cryogenic ltd ) . here and all throughout the paper , we have plotted the measured magnetic moment in cgs units ( corresponding to a total maghemite volume @xmath7 ) . zfc " curve is measured in the usual way by cooling the sample ( down to 10 k ) in zero field , applying a field and then raising the temperature ; the field - cooled one ( fc " ) is obtained while cooling in the field @xmath8 . the zfc curve shows a broad maximum around @xmath9 . it represents the progressive deblocking of larger and larger particles as the temperature t is raised . let us consider that a particle of volume @xmath10 involves an anisotropy barrier @xmath11 , where @xmath12 is a density of anisotropy energy . if the time spent at a given t is @xmath13 ( @xmath14 ) , then for thermally activated dynamics most particles which are being deblocked at t have a typical volume v obeying an arrhenius law @xmath15 where @xmath16 is a microscopic attempt time @xcite . by assuming in addition that the saturated moment of a particle is proportional to its volume , and that the moments follow a langevin function when they are deblocked ( superparamagnetism ) , we have calculated the zfc curve corresponding to the measured size distribution . the peak is obtained at the measured temperature for @xmath17 . this value is in agreement @xcite with high - field measurements where the integral of the work needed for saturating the sample has been evaluated and also with mssbauer spectroscopy results . it is one order of magnitude larger than the bulk maghemite value , as commonly observed in small particles where shape and surface contributions have increased the magnetic anisotropy @xcite . note that , due to the distribution width and to the 1/t variation of superparamagnetism , the zfc - peak is found at a temperature which is three times larger than that corresponding to the peak value @xmath18 of the size distribution ( @xmath19 ) @xcite . the setup used for the low - t experiments is a home made combination of an r.f . squid magnetometer @xcite and a dilution refrigerator @xcite . the sample is coupled to the mixing chamber through a thermal impedance which allows a temperature range of 35 mk to 7 k. for relaxation measurements at the lowest temperatures , some spurious heating has been found when the field is varied , due to eddy currents in the thermalization link ; we have therefore carefully adjusted the field amplitude , and chosen a `` slow '' cut - off procedure ( 5 s ) , in such a way that the results become independent of both these parameters . we also have limited our lower range to 60 mk . the sample is first cooled in zero field from room temperature to the dilution regime . from that point , the temperature can no longer be easily raised above 7 k. the procedure for the relaxation measurements at @xmath20 starts with heating the sample to a high enough temperature for deblocking of all particles which may participate in the dynamics at @xmath21 , e.g. 7 k. then the sample is field - cooled from 7 k to @xmath21 , the field is decreased to zero and the squid signal variation corresponding to the slow relaxation processes is measured . this procedure of not heating up to room temperature makes sense because our sample is highly diluted ; in a first approximation the particles can be considered independent of each other . we have checked that our choice of the reinitialization temperature had no influence on the resulting dynamics . figure 2 presents examples of relaxation curves . they are roughly logarithmic in time , apart from some uncertainty in the first seconds , which should be related to the 5 s field cut - off duration . in this paper , we only consider the average logarithmic slope of the curves ( magnetic viscosity " ) , which we determine between @xmath22 and @xmath23 . figure 3 shows our set of results . for decreasing temperatures , the measured viscosity first decreases , then flattens out , and surprisingly increases back below 150 mk . we present a simple model for the t - dependence of the viscosity before discussing this result in more detail . by thinking of the sample relaxation at t as a sum of independent processes , one may write the total relaxing moment @xmath24 as @xmath25 where the summation runs over the barrier distribution p(u ) associated with the size distribution of the particles . @xmath26 stands for the field - cooled moment " of the particles with anisotropy barrier u , which is the thermal average of the moments at their blocking temperature ; as a first approximation , one may assume @xmath11 and @xmath27 , hence @xmath28 . at any temperature @xmath29 and after a time @xmath13 following the field cut - off , one may consider that the only relaxing objects are those for which @xmath30 . the logarithmic derivative @xmath31 of the magnetization ( magnetic viscosity ) can then be easily derived as @xmath32 the magnetic viscosity is commonly expected to be proportional to t @xcite , a controversial point since in our cases of interest the energy barrier distribution @xmath33 may vary significantly @xcite,@xcite . indeed , from eq . [ s1 ] , one sees that the distribution of interest is @xmath34 rather than @xmath33 itself ; with @xmath35 , eq . [ s1 ] then becomes @xmath36 we believe that these @xmath13 and @xmath37-dependences of the viscosity are probably hidden in most experimental results , due to the combination of the distributions @xmath33 and @xmath26 which are not accurately known ( the @xmath38-variation of the magnetization is very close to @xmath39 , due to the microscopic value of @xmath40 ) . however , it seems to us that the first approximation of the viscosity in the case of non - interacting particles with a flat distribution of barriers should be a quadratic rather than a linear function of temperature . as expected from thermally activated dynamics and a regular distribution of barriers , the 0.5 - 5 k viscosity is seen to decrease for decreasing temperatures . it shows a slight upwards curvature which is compatible with a @xmath37-dependence and a flat distribution ; actually , this t - range corresponds to the blocking of 2 - 3 nm objects , which are not well characterized from the distribution in fig.1 . however , it is clear from fig.3 that a normal extrapolation will not yield a zero viscosity at zero temperature ; below 150 mk , the viscosity data even show a systematic tendency to increase as t is lowered . a similar behavior has been noted in an array of cobalt particles @xcite , and also in a permalloy sample @xcite . with respect to maghemite , a viscosity anomaly ( plateau from 2.2 to 1.8 k ) has been observed in a system of particles dispersed in a glassy matrix @xcite ; no anomaly was visible for the same particles in water , suggesting the influence of the matrix via magnetostriction phenomena @xcite . we consider that our present results may give rise to two possible conclusions ( a combination of both is also possible ) . first , one may assume that the dynamics is thermally activated . the implication of our results is that the distribution of energy barriers @xmath33 increases abruptly towards smaller values , more rapidly than @xmath41 . this is a surprising result , very different from the framework in which viscosity measurements are commonly interpreted in the literature ( approximately a flat distribution ) . we have in addition performed zfc / fc measurements in this low - t range , which are displayed in fig.4 . they show an increase in the magnetization for decreasing t , which is 1/t - like and of the same amount in both zfc and fc cases ( see trm in fig . 4 ) . if this behavior is ascribed to clusters of e.g. 10 spins , the curie constant would correspond to 0.5% of the total amount . thus , there are indeed some very small magnetic entities which are not frozen , even at 60 mk . 4 also shows a significant difference between the zfc and fc curves , which corresponds to the slow dynamics observed in this low - t range . all data are therefore compatible with the existence of a significant low - energy tail of the barrier distribution , increasing further for the lowest values . one may think of very small particles ; it would be of interest to check other systems of small particles for this possibility . it has also been proposed that such small barriers arise from decompensation effects at the surface of the ferrimagnetic particles @xcite ; surface defects might be an intrinsic component of the dynamics of nanometric particles at very low temperatures . a second possible conclusion concerns the quantum tunneling of the particle magnetization ( qtm ) through its anisotropy barrier . in a first approximation , the contribution of such processes could be independent of temperature ; from @xcite , quantum processes should be of the same order of magnitude as thermal processes below a crossover temperature @xmath42 , which can be here estimated as @xmath43 ( @xmath42 does not depend on the barrier height , which only influences the relaxation rates ) . it is therefore possible that such processes contribute significantly in our t - range ( one may even wonder why they should not be visible ) . the increase of the viscosity towards lower t can be understood in two ways . on the one hand , it has been argued in @xcite that the viscosity should be t - independent if the two energy levels between which quantum tunneling occurs are sufficiently separated with respect to @xmath44 , whereas it should go like 1/t for quasi - degenerate levels , which could be our situation of low - field relaxations . a low - t increase of the viscosity in permalloy has thus been described as quantum jumps of a bloch wall between pinning sites of comparable energies @xcite . in more general terms , on the other hand , one may think that lower temperatures decrease the coupling to phonons , therefore reducing the dissipation and enhancing quantum tunneling processes @xcite . a t.lnt " plot has been proposed to help distinguish between thermal and quantum processes in size - distributed particles @xcite , but this is not possible with the present relaxation data , obtained by measuring only squid signal variation ( and not the full value of the magnetization ) . actually , the question of a satisfactory evidence of qtm processes in such systems remains controversial ; however , we believe that the numerous observations of anomalies in the low - t dynamics of small particles lead us to the minimal conclusion that things are not as simple as we had thought .	the magnetization relaxation rate of small particles dispersed in a silica matrix has been measured from 60 mk to 5 k. it shows a minimum around 150 mk , that can be discussed in terms of either thermal or quantum relaxation regime .
the next generation of large - area survey facilities ( e.g. , pan - starrs , lsst , des , vista / vst ; @xcite ) will image orders of magnitude more objects than it will be possible to obtain spectra for . as a result , robust determination of redshifts from photometric data is of crucial importance to these projects . while the distinctive `` 4000@xmath4break '' facilitates photometric redshifts ( `` photo-@xmath5 s '' ) for galaxies ( e.g. , * ? ? ? * ; * ? ? ? * ) , the majority of quasars ( @xmath6 ) lack such a strong spectral feature at observed - frame optical wavelengths , making the determination of photometric redshifts for low-@xmath5 quasars more challenging . while it has been proven to be possible to determine photo-@xmath5 s of quasars from the changes that their emission lines induce in their broad - band photometry ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and from spectral energy distribution ( sed ) fitting ( e.g. , * ? ? ? * ; * ? ? ? * ) , the accuracy of these photo - z s does not match that of galaxies and often suffers from catastrophic failures ( @xmath7 ) . here we introduce a powerful new tool to aid redshift determination for emission - line objects ( such as quasars ) by taking advantage of the spectroscopic properties of the earth s atmosphere . light rays from extraterrestrial sources are bent according to snell s law as they enter the earth s atmosphere from the vacuum of space . as a result , except at the zenith , a celestial source observed from the earth will appear higher in the sky than it actually is . the magnitude of this deflection depends on the index of refraction in air and the photon s angle of incidence . since the index of refraction of air is a function of wavelength , light rays passing through the atmosphere undergo dispersion , whereby shorter wavelength photons are bent more than longer wavelength photons . @xcite provides an in - depth discussion of this effect , reminding observers of the rather large magnitude of this effect even at moderate airmasses and the need to perform spectroscopy at the `` parallactic angle '' . this effect , known as differential chromatic refraction ( dcr ; e.g , @xcite ) , is normally a source of nuisance for multi - band astrometry . indeed oftentimes additional lenses are used to form an atmospheric dispersion corrector ( adc ) to compensate for dcr in hardware , such as for the 2df system @xcite on the anglo - australian telescope . however , rather than being a nuisance , dcr can be used as a unique tool for estimating redshifts of emission - line objects . software - based astrometric corrections for dcr are generally computed as a function of broad - band flux ratio ( i.e. , color ) , whereas the actual dcr in emission - line objects instead depends on the distribution of flux _ within _ the bandpass . the difference between the expected and observed astrometric displacements due to dcr enables determination of _ astrometric redshifts _ ( or astro-@xmath5 s ) for strong emission - line objects . this discrepancy can be used separately from , or in conjunction with , more traditional photo-@xmath5 estimates to provide redshifts from imaging data . indeed , contrary to conventional wisdom , the next generation of multi - epoch imaging surveys may want to perform a fraction of their observations at moderately high airmass in order to take full advantage of this effect . the sections of this paper are as follows . section 2 discusses dcr from a theoretical perspective , while 3 considers dcr observations from both single- and multi - epoch quasar data from the sloan digital sky survey ( sdss ; @xcite ) . section 4 examines how these measurements can be used to improve photometric redshift estimation for quasars . finally , 5 discusses some avenues for further work and 6 presents our conclusions . differential chromatic refraction is a fact of life for any multi - band astronomical imaging survey . because of the wavelength dependence of the refractive index of air , a source observed with a blue filter will appear slightly higher in the sky than the same source observed through a red filter . figure [ fig : dcr ] illustrates this effect . the left - hand panel indicates the magnitude of the effect as a function of wavelength ; here for the sdss filter set . the right - hand panel of figure [ fig : dcr ] illustrates why the blue image is higher in the sky than the red image . it is clear that refraction is a potentially large source of astrometric error and both it and dcr must be carefully corrected ( either in hardware with an adc or in software ) in order to produce accurate astrometric solutions . here we explicitly consider the absolute theoretical wavelength depedence of the deflection of photons from astronomical sources . the angular deflection of an incoming photon by the earth s atmosphere is given by @xmath8 where @xmath9 is the angle from the zenith ( airmass , @xmath10 $ ] ; good for @xmath11 ) and @xmath12 is determined from the index of refraction , @xmath13 , as @xmath14 ( e.g. , * ? ? ? the wavelength dependence of the index of refraction of air can be described by @xmath1510 ^ 6=64.328 + \frac{29498.1}{146-(1/\lambda)^2 } + \frac{255.4}{41-(1/\lambda)^2},\label{equ : n_lambda } \end{aligned}\ ] ] where @xmath16 is expressed in microns ( e.g. , * ? ? ? thus , a flat - spectrum source , where a flat - spectrum source has @xmath17 . ] observed at am=1.414 ( @xmath18 ) in the sdss @xmath19 band ( 6165 ) will appear @xmath20 higher in the sky ( ignoring temperature , pressure , water vapor , and altitude effects ) than it would in the absence of earth s atmosphere ; see figure [ fig : dcr ] . to determine the dcr for any given object , it is sufficient to know the effective wavelength of the object within a given bandpass . following @xcite , effective wavelengths are given by @xmath21 the subscript @xmath22 represents a given bandpass , @xmath23 or @xmath5 @xcite . the incident flux at a given wavelength and redshift is indicated by @xmath24 while @xmath25 is the response of the bandpasses . the transmission curves , @xmath25 , for the sdss system include atmospheric extinction at the mean airmass of the sdss survey , @xmath26 . for a flat - spectrum source , the nominal effective wavelengths of the sdss bandpasses are 3551 , 4686 , 6165 , 7481 , and 8931 , respectively for @xmath27 , @xmath28 , @xmath19 , @xmath29 , and @xmath5 . for a power - law continuum with spectral index @xmath30 , characteristic of quasars , the effective wavelengths are 3541 , 4653 , 6147 , 7461 , and 8904 , respectively @xcite . using these effective wavelengths , the dcr can be determined from equations 13 . in reality , the effective wavelength will depend on the more complicated wavelength dependence for the seds of real astronomical sources . as broad - band photometry trades a loss of information ( integration of a spectral energy distribution over @xmath31 ) for a gain in signal - to - noise per unit time , the information from spectral features narrower than the photometric bandpasses is marginalized . quasar emission lines , however , are sufficiently broad and strong that their effects are quite noticeable both on their colors ( e.g. , @xcite ) , and their broad - band magnitudes @xcite . these emission lines can similarly shift the effective wavelength of a photometric bandpass . this shift occurs because the effective wavelength depends not only on the filter s transmission properties , but also on the distribution of the source s flux within the bandpass . indeed , while astronomers are frequently tempted to assume a uniform transformation between broad - band photometric systems , those transformations are , in fact , strongly dependent on the ( observed frame ) spectral energy distribution of the objects in question ( e.g. , * ; * ? ? ? * ) . in figure [ fig : examples ] we illustrate the effect of quasar emission lines on the effective wavelengths within the sdss broad - band photometric system . shown is the @xcite sdss composite quasar spectrum at four redshifts where the ly@xmath32 or emission lines lie at the red or blue edge of the @xmath27 filter . figure [ fig : lameff ] shows the full redshift dependence of the effective wavelength for the mean quasar ; here we have convolved the mean quasar spectrum with the sdss bandpasses according to equation [ eq : lameff ] . compared to a power - law continuum source , the effective wavelength of the sdss bandpasses can change by as much as 150 for quasars . ( for @xmath33 and @xmath34 the @xcite composite spectrum does not fully span the @xmath5 and @xmath27 bands , respectively and the expected changes in effective wavelength can not be determined from the composite quasar spectrum . ) we can use these effective wavelengths to determine the magnitude of the dcr for emission - line objects as a function of redshift , but first we need to understand how dcr corrections are applied , particularly in the case of sdss astrometry . as can be seen from figure [ fig : dcr ] , if no dcr corrections were applied , then potentially large positional differences between bandpasses would severly complicate multi - color photometry . since the true sed of an object is unknown , the broad - band photometry itself provides the best model for the wavelength dependence of the sed , which is needed to determine the dcr correction . in practice , this does not involve making a full correction based on a theoretical model , but rather by including a color - dependent term in the astrometric solution ( e.g. , * ? ? ? * ) . for sdss imaging , the dcr correction is modeled as a linear function of color for stars bluer than @xmath35 in the @xmath27-band and @xmath36 in the @xmath28-band ; redder stars are modeled with constant color . for the @xmath19 , @xmath29 , and @xmath5 bands , the dcr is taken to be a linear function of the @xmath37 color for all stars . for more details , see 5.3 in @xcite . this color - dependent modeling of the dcr means that , for purely power - law sources , the positional agreement between the five astrometrically - calibrated sdss bandpasses should not show any evidence of dcr . any residual astrometric offsets will be due to the difference between the color - dependent correction that is applied to the sdss astrometry and the actual dcr for any particular object , which depends on the distribution of flux _ within _ a bandpass rather than the distribution of flux _ across _ bandpasses ( i.e. , the color ) . as such , for our purposes , we can ignore any altitude , temperature , pressure , or water vapor corrections to the index of refraction of air since we are only concerned with the _ relative _ dcr with respect to a presumed model . as a result of the broad - band color - dependent dcr correction , we can not simply use the effective wavelengths from the composite quasar spectrum to determine the expected positional offsets of quasars . rather , we must compare these offsets with those expected from the astrometric model . for our analysis , we will consider two models for dcr . one is based on synthetic colors determined from the composite quasar spectrum ( e.g. , * ? ? ? * ) , the other is based on the mean power - law sed of quasars , @xmath38 . accordingly , we modify equation [ eq : r ] to @xmath39\tan(z)\label{eq : rnew2}\ ] ] or @xmath40\tan(z).\label{eq : rnew}\ ] ] once we have determined @xmath41 ( and thus @xmath12 ) for both the composite spectrum and the model in each bandpass as a function of redshift , positional offsets can be calculated for any airmass according using equation [ eq : rnew2 ] . figure [ fig : offsetband ] shows the predicted offsets for the mean quasar as a function of redshift and airmass in each of the five sdss bandpasses . here we make no airmass corrections to the sdss bandpasses despite the fact that they are given specifically for @xmath26 . as demonstrated by d. schlegel ( unpublished ) , the 25 - 35 mas rms astrometric errors between bandpasses of the sdss imaging survey @xcite are comparable to or less than the astrometric offsets due to quasar emission lines redshifting through the sdss broad - band filters . looking ahead to comparisons with the data , it must be realized that the dcr effect is entirely in the direction along the `` parallactic angle '' ( or perpendicular to the horizon ) , thus we adopt the notation @xmath42 in figure [ fig : offsetband ] to indicate positional offsets in the direction where dcr is applicable . our sign convention is that bluer effective wavelengths result in images higher in the sky than expected , yielding positive astrometric offsets . we will see in [ sec : data ] that the positions of sdss objects in the @xmath43 bands are given with respect to the @xmath19 band , which defines the sdss astrometric system . that is the @xmath19-band positional offsets do not agree with figure [ fig : offsetband ] , but rather are all identically zero ( by definition ) . thus it will often be more convenient to examine the relative changes in position between two bandpasses than the absolute deviation for a single bandpass . the difference in positional offsets between two adjacent colors , @xmath44 is just @xmath45 , where @xmath46 and @xmath13 represent the two bandpasses and . figure [ fig : offsetcolor ] shows the resulting @xmath47 relations as a function of redshift for adjacent bandpasses at various airmasses . at sufficiently high airmass or by averaging multi - epoch observations at low airmass , these positional offsets can exceed the astrometric errors and can be used as quasar redshift diagnostics . while the dcr effect is a well - known phenomenon in astrometry and is generic to all ground - based data sets , our application of it to astrometric redshifts is based on data from the sloan digital sky survey ( sdss ) . the sdss photometric data is summarized in a series of papers including @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite . in [ sec : multi ] , we will make particular use of the data from sdss `` stripe 82 '' @xcite , where multiple epochs of sdss imaging data have been combined , reducing both the photometric and astrometric errors . in addition , we will make use of spectroscopically - confirmed sdss quasars that were selected according to @xcite and tiled according to @xcite . discussions of the sdss filter system and astrometric solutions can be found in @xcite , @xcite , @xcite , and also at http://www.sdss.org/dr7/algorithms/astrometry.html . we now compare our theoretical predictions with actual measurements for the 77,000 quasars from the dr5 quasar catalog @xcite . sdss astrometry is reported with respect to the @xmath19 band ( corrected to am=1 ) . the positions of the quasars in the other bandpasses were extracted from the photoobjall in the sdss database . these values are given in the database as offsetra@xmath48 ( = ra@xmath48-ra@xmath49cos[dec@xmath49 ] ) and offsetdec@xmath48 ( = dec@xmath48-dec@xmath49 ) , where @xmath51 represents one of the @xmath52 bandpasses . by definition offsetra@xmath49 and offsetdec@xmath49 are identically zero as the @xmath19 band ( corrected for atmospheric refraction ) defines the sdss astrometric system . the dr5 quasar catalog spans a redshift range of 0.08@[email protected] and an airmass range of 1.002@[email protected] . as we are interested only in the positional offsets due to dcr , we must first determine the parallactic angle for each of the observations and project the observed offsets onto it . the parallactic angle , @xmath54 , is given by @xmath55 where @xmath56 is the latitude of the observatory and @xmath57 is the `` hour angle '' of the observation ( e.g. , * ? ? ? that portion of the positional offsets parallel to the parallactic angle comes from a combination of astrometric errors and atmospheric dispersion , while the portion perpendicular to the parallactic angle is due only to astrometric errors . figure [ fig : offsetplot2 ] shows the resulting @xmath47 for the sdss @xmath58 bandpasses . while the deviation of the astrometric offsets from the model is more noticeable at higher airmass and most sdss observations are performed at the minimum possible airmass , the effect is still quite apparent in the ensemble average . although figure [ fig : offsetplot2 ] shows that emission - line - induced astrometric offsets for quasars are evident in single - epoch sdss data , they are not significantly larger than the mean astrometric errors for individual observations , reducing the utility of the astrometric offsets for redshift determination . however , for the sdss southern equatorial region ( aka , `` stripe 82 '' ) and for the upcoming multi - epoch synoptic surveys ( e.g. , pan - starrs and lsst ) , it is possible to combine data from different epochs in order to reduce random astrometric errors and increase the signal - to - noise of the astrometric offsets along the parallactic angle . to illustrate this improvement we have queried the sdss stripe 82 database for multiple observations of objects . while stripe 82 has dozens of repeat scans , not all of these observations were observed at airmass high enough to show the dcr effect and not all were observed at the same airmass . thus , we only used observations objects with @xmath59 and we have combined multiple such observations together , scaling by @xmath60 to a mean airmass of 1.3 . this approach allows us to show not only the benefits of combining the few dozen epochs available from sdss stripe 82 data , but also the hundreds of epochs that will be available in future imaging surveys . figure [ fig : qozr ] shows the resulting , more accurate , positional offsets in the @xmath27 , and @xmath28 bandpasses , along with their difference for 6430 quasars in sdss stripe 82 . comparison of the bottom panel of figure [ fig : qozr ] and the top panel of figure [ fig : offsetplot2 ] demonstrates that combining multiple observations has significantly reduced the positional scatter . the range of astrometric offsets is nearly 300 mas ( e.g. , @xmath61 vs. @xmath62 ) , as compared with the 2535 mas astrometric errors for the single - epoch sdss data . while few objects have offsets considerably larger than 30 mas , over 60% of the quasars have astrometric offsets at least this large . moreover , by reducing random scatter with combined multi - epoch data , astrometric offsets smaller than 30 mas carry useful information . for example a measured value of @xmath63 may not uniquely identify the redshift , but it does suggest redshift ranges that are higly unlikely . figure [ fig : nvsdx ] illustrates the reduction of random astrometric errors from 2535 mas in single epoch observations to under 10 mas after a few dozen combined observations . to determine the astrometric improvement we took the @xmath27 and @xmath28 band offsets perpendicular to the parallactic angle for 50 hot white dwarf observations that is we looked at the astrometric offsets in the direction where we expect only random errors and not systematic dcr effects . those observations were bootstrap resampled leaving out n objects at a time from 1 to 50 in order to determine the mean improvement as a function of the number of epochs combined . while individual observations are noisy , the mean offets in the direction parallel to the parallactic angle ( due to both signal and noise ) are much larger than the perpendicular scatter ( due to noise only ) . these differences bode well for the inclusion of astrometric offsets in photometric redshift estimation and , perhaps , in object classification ( e.g. , see the object classification work by @xcite based on proper motions of objects in stripe 82 ) . figure [ fig : dispmulti ] further illustrates the effect of combining multiple observations . here we compare both the expected and observed positional offsets for a hot white dwarf , a @xmath64 quasar , and a @xmath65 quasar at @xmath66 . while the individual observations are noisy , the mean positions are in good agreement with our predictions . more importantly , the offets in the direction parallel to the parallactic angle are much larger than the perpendicular scatter and , for different objects and different redshifts the parallel offsets are noticeably distinct . these differences bode well for their use in photometric redshifts and perhaps in object classification ( e.g. , see the object classification work by @xcite based on proper motions of objects in stripe 82 ) . since the random astrometric errors for the combined multi - epoch data are less than the range of astrometric offsets for quasars , these values can aid in redshift estimation . figure [ fig : qczr ] illustrates how dcr can be helpful in this endeavor . for example , emission features cause a blueward dip in the @xmath67 color for quasars with @xmath68 . the @xmath67 color is roughly degenerate for a quasar at @xmath64 and @xmath69 , which is a source of catastrophic error for normal photo-@xmath5 s . the color - coding of figure [ fig : qczr ] shows that including the @xmath67 astrometric offset breaks this degeracy by identifying the color of the @xmath64 quasars as being due to the emission feature on the blue side of the bandpass rather than on the red side of the bandpass . because the astrometric offsets are roughly gaussian distributed ( as are quasar colors ) , it is straightforward to include such measurements in our existing algorithm for redshift estimation as discussed in @xcite . we can simply treat the observed @xmath70 and @xmath71 measurements as additional `` colors '' . this method compares each observed color to the mean ( and standard deviation ) color at every redshift , choosing the redshift that minimizes the @xmath72 . for the 6430 quasars from stripe 82 we have been able to combine multiple observations , reducing both the photometric and astrometric errors . to determine the photo-@xmath5 improvement due to our astro-@xmath5 s , we must first determine the photo-@xmath5 distribution using the improved photometry of these objects since we want to distinguish between astrometric and photometric improvements . s for these 6430 quasars are first computed using only the five sdss magnitudes ( and errors ) . we perform 5-fold cross - validation whereby 80% of the sample is used for training and 20% is used for testing . five iterations are used so that photo-@xmath5 s are estimated for the full sample . these 4-color photo-@xmath5 s provide a baseline for comparison with redshift estimates that also include astrometric offsets . we then repeat this process , now adding the @xmath70 and @xmath71 measurements , treating them as the 5th and 6th colors . any improvement over the 4-color photo-@xmath5 s must be due to the added astrometric information . figure [ fig : zz ] compares the spectropic versus photometric ( and photometric+astrometric ) redshifts for the full sample . objects appearing along the diagonal have accruately estimated redshifts , whereas off - diagonal objects indicate catastrophic photometric redshift errors @xcite . while there are still errors in the photometric+astrometric sample , it can be seen that inclusion of astrometric information results in somewhat fewer catastropic errors and a tighter distribution along the diagonal ( i.e. , the precision of those objects with accurate photometric redshifts has been improved ) . the improvement can be better seen when plotting the fractional redshift error as shown in figure [ fig : dzhist ] . adding astrometric redshifts reduces the fractional error from 5% to 3% ( excluding catastrophic errors ) . it also increases the fraction of objects within @xmath73 by 3% and the fraction within @xmath74 by 9% . we emphasize that this improvement comes simply from treating the residual astrometric offsets as colors using our existing photo-@xmath5 algorithm . in reality , it would make more sense to use the offsets in a different manner . in particular , in addition to reducing the scatter in quasar redshift estimates , figure [ fig : qczr ] suggests that astro-@xmath5 s can help break color degeneracies and will help improve redshift estimates for objects with catastropic errors . for example , broad absorption line quasars ( e.g. , * ? ? ? * ) and dust reddened quasars @xcite can ruin color - based photometric redshifts algorithms . high - redshift quasars become very red in @xmath67 as a result of lyman-@xmath32 forest absorption , but intrinsically blue quasars that are reddened by dust can appear just as red , which results in catastrophic photo-@xmath5 errors . however , the atmospheric dispersion effect only depends on the distribution of flux _ within _ the bandpass . as such , a moderate amount of dust has little effect on the positional offsets due to strong emission lines . this suggests that astro-@xmath5 s may perform better when used as priors in advance of a standard photo-@xmath5 algorithm . we have concentrated on the applications of astrometric offsets specifically for quasars ; however , the method is potentially applicable to other emission - line objects ( e.g. , type 2 quasars and supernovae ) . for type 2 quasars the emission lines are narrow , but their strength means that we can expect to see measurable astrometric offsets that are a function of redshift . using a template type 2 quasar spectrum ( e.g. , * ? ? ? * ) , we can estimate the astrometric offsets due to strong emission lines in type 2 quasar spectra . figure [ fig : offsetcolortype2 ] shows that , over the range @xmath75 , @xmath76 falls by almost 90 mas , while @xmath76 rises by 75 mas . thus redshifts for type 2 quasars with @xmath77 could be distinguished at roughly the same level of accuracy as type 1 quasars . it may also be possible to estimate redshifts for supernovae using astrometric offsets . while their seds are highly time - dependent and it would not be possible to combine multiple epochs , if the distinctive emission features at each epoch produced redshifts estimates that were in agreement , these astro-@xmath5 s might be as reliable as they would be if the seds were not time - dependent . most sne spectral templates do nt cover enough of the uv to properly estimate their positional offsets with redshift , but based on the existing coverage longward of 3000 , it would appear that the range of @xmath27- and @xmath28-band offsets is as large as @xmath78certainly enough to make an initial guess at the redshift . just as the nature of the atmospheric offsets allows improvements in redshift estimation , the offsets could also be utilized in object classification itself . while astrometric solutions attempt to remove dcr , they can only do so at the level of broad - band photometric accuracy , whereas dcr is sensitive to features smaller than the width of the photometric bandpasses . as such , objects with distinctive seds may stand out enough in astrometric offset space to aid in object classification . while astrometric offsets alone are not likely to be useful for object classification , coupling these astrometric measurements with color and variability information within the context of modern bayesian classification algorithms @xcite could prove very fruitful in terms of the accuracy of object classification from broad - band imaging data . while most surveys aim to obtain observations as close to the meridian as possible ( and thus at the lowest airmass possible ) , any object with declination different from the latitude of the observatory will necessarily be observed at @xmath79 . thus any imaging survey will naturally yield observations that can be used with our approach . as refraction increases with airmass , one should ask whether future surveys may wish to obtain a fraction of their observations at relatively high airmass . such observations will result in a loss of photons ( particularly in the ultraviolet ) , so one must consider the details of the tradeoff between low and high airmass observations . it begs the question of whether future surveys might actually want to go out of their way to make some observations at high airmass . such observations will also result in a loss of photons and will complicate data reduction , so it is logical to consider the trade - off between a smaller number of high airmass observations and a larger number of low airmass observations . in both cases one seeks to maximize the signal - to - noise of the effect by either making the noise smaller or the signal larger . we have already shown in figure [ fig : nvsdx ] that combining just a few observations can go a long way towards reducing random astrometric errors below the level of anomalous dcr offsets for quasars . we next ask whether there is utility in making observations at very high airmass in order to increase the snr of the anomalous dcr effect . in figure [ fig : amtest ] we overplot the expected offsets for quasars in both the @xmath27- and @xmath28-bands for airmasses 1.1 , 1.4 , and 1.8 . it can be seen that an airmass of 1.1 is not sufficient for breaking redshift degeneracies , which implies that some level of concerted effort will be needed to obtain observations off of the meridian . at an airmass of 1.4 , on the other hand , there is sufficient range to clearly indentify certain redshifts , even considering the scatter of the measurements as compared to these theoretical distributions . extending to @xmath80 , however , seems to offer little improvement , perhaps with the exception of the @xmath81 and @xmath82 . a full cost - benefit analysis of the trade - offs between more observations and higher airmass is beyond the scope of this paper as such analysis is highly system dependent , but it would seem that of order 10 observations at a moderate airmass of @xmath83 would be sufficient to yield significant improvement in photometric redshifts for quasars without causing significant problems for observational cadence or data reduction . while the gains in photometric redshift accuracy due to the addition of astrometric information shown herein are modest ( @xmath849% ) , they undersell the utility of astrometric redshifts for the next generation of imaging surveys which will be able to combine hundreds instead of tens of epochs of data , each epoch with astrometric accuracy superior to that from the sdss . for example , at @xmath85 the lsst astrometric errors are expected to be @xmath8415 mas and there will exist over 200 epochs of data @xcite . on the other hand , figure [ fig : amoffsets ] shows the increase of positional offsets with airmass for @xmath27 vs. @xmath28 and @xmath28 vs. @xmath19 as a function of quasar redshift . here we overplot @xmath86 and @xmath87 to illustrate that the lack of signal in one measurement ( e.g. , @xmath63 does not constrain the redshift ) can be offset by a strong signal for the other measurement . the combination of figure [ fig : amoffsets ] and figure [ fig : nvsdx ] shows that we expect to see a measurable signal even at relatively low airmass and/or relatively small number of observations , but that either increasing the airmass or increasing the number of observations significantly improves the signal - to - noise . herein we have merely included astrometric offsets in our existing photometric redshift algorithm . however , more judicious use of the astrometric information should enable further gains in redshift estimation , particularly in terms of identifying objects that are likely to have catastrophic photometric redshift estimates . throughout this work we have ignored temperature and pressure differences between observations , but in the future , our results may benefit from making corrections for temperature and pressure . we have assumed a pressure of @xmath88pa and a temperature of @xmath89k . positional offsets at other temperatures ( t ) and pressures ( p ) can scale according to @xmath90 nighttime temperatures range between about -15 c and + 15 c , thus temperature variations may cause up to 12% error in combining multi - epoch data . however , the stripe 82 data is not taken over the course of the full year , so the actual temperature variations are much smaller and we are probably justified in ignoring this effect . more important than temperature and pressure differences are differences in airmass . by the nature of our combining multi - epoch observations , we must also combine observations at different airmasses . for now we have restricted ourselves to combining observations in a small range of airmass . however , the astrometric offsets scale with airmass according to @xmath91 as a result , there is really no reason that we ca nt combine observations over a broader range of airmasses . astrometrists have long been dealing with the dispersive nature of our atmosphere when performing broad - band photometry and astrometry . here we show that the dispersive nature of our atmosphere can actually be a boon rather than a burden . objects with strong emission lines produce measurable astrometric offsets ( with respect to an expected model ) in the ensemble average even when observed at relatively low airmass . the signal - to - noise of the effect can be maximized by either combining many observations at low airmass or fewer observations at higher airmass . with well - measured astrometric offsets , it is possible to both break degeneracies in photometric redshift estimates of quasars and to reduce their overall scatter . for combined , multi - epoch data from the sdss stripe 82 region , adding astrometric information increases the fraction of quasars with correct photometric redshifts by as much as 9% ( @xmath92 ) . given the lack of spectroscopic capabilities for future imaging surveys , the community would do well to build astrometric redshifts into their survey planning . gtr acknowledges support from an alfred p. sloan research fellowship . we thank eljko ivezi , jeff munn , and don schneider for comments on the manuscript and bob hindsley for early work on dcr in the sdss astrometric system . funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s . department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england . the sdss is managed by the astrophysical research consortium for the participating institutions . the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . , n. , et al . 2002 , in survey and other telescope technologies and discoveries . edited by tyson , j. anthony ; wolff , sidney . proceedings of the spie , volume 4836 , pp . 154 - 164 ( 2002 ) . , j. a. tyson & s. wolff , vol . 4836 of _ presented at the society of photo - optical instrumentation engineers ( spie ) conference _ , 154164	the wavelength dependence of atmospheric refraction causes differential chromatic refraction ( dcr ) , whereby objects imaged at different optical / uv wavelengths are observed at slightly different positions in the plane of the detector . strong spectral features induce changes in the effective wavelengths of broad - band filters that are capable of producing significant positional offsets with respect to standard dcr corrections . we examine such offsets for broad - emission - line ( type 1 ) quasars from the sloan digital sky survey ( sdss ) spanning @xmath0 and an airmass range of 1.0 to 1.8 . these offsets are in good agreement with those predicted by convolving a composite quasar spectrum with the sdss bandpasses as a function of redshift and airmass . this astrometric information can be used to break degeneracies in photometric redshifts of quasars ( or other emission - line sources ) and , for extreme cases , may be suitable for determining `` astrometric redshifts '' . on the sdss s southern equatorial stripe , where it is possible to average many multi - epoch measurements , more than 60% of quasars have emission - line - induced astrometric offsets larger than the sdss s relative astrometric errors of 25 - 35 mas . folding these astrometric offsets into photometric redshift estimates yields an improvement of 9% within @xmath1 . future multi - epoch synoptic surveys such as lsst and pan - starrs could benefit from intentionally making @xmath2 observations at relatively high airmass ( @xmath3 ) in order to improve their photometric redshifts for quasars .
kondo insulators as well as heavy fermion materials @xcite are believed to be described by kondo lattice models with antiferromagnetic exchange @xmath0 and hopping @xmath1 . in this framework , the kondo insulator - exemplified by ce@xmath4bi@xmath5pt@xmath4 - corresponds to a special band filling in which there is exactly one conduction electron per impurity spin . the origin of the charge and spin gaps is easy to understand in the limit of large @xmath6 where each conduction electron is trapped by an impurity spin to form a kondo singlet . starting from this point there are two ways of generating metallic states which will described the heavy electron state . on one hand , one can remove conduction electrons thus leaving uncompensated impurity spins . in the limit @xmath7 the problem maps onto the @xmath8 hubbard model where the kondo singlets are represented by empty sites and the bachelor - or unscreened - spins by electrons @xcite . thus , a metallic state is expected . on the other hand one can keep the number of conduction electrons constant and deplete the lattice of impurity spins . starting from ce@xmath4bi@xmath5pt@xmath4 this amounts to replacing ce by la @xcite . each missing impurity spin liberates a conduction electron . as can be seen easily in the strong coupling limit this conduction electron is localized since its motion involves the breaking of adjacent kondo singlets . hence a bound state with magnetic properties appears within the gap @xcite . if the depletion is _ large _ enough the induced localized levels may overlap coherently to form a metallic state . alternatively this metallic state where there are more conduction electrons than impurity spins may be viewed as evolving from dilute impurity limit as realized for example in ce@xmath9la@xmath10cu@xmath11 @xcite . note however that in the small ( large ) @xmath12 limit the ce ( la ) atoms are randomly substituted for la ( ce ) in the lacu@xmath11 ( cecu@xmath11 ) crystal structure . in the next section , we first consider within a mean - field approximation @xcite both above describe routes to obtain metallic states . the paramagnetic solution of this mean - field approximation is nothing but the large - n ( n corresponds to the su(n ) symmetry of the impurity spins ) saddle point @xcite . here , we concentrate on a special choice of depletion scheme for the localized spins . three out of four impurity spins are removed in a regular way thus ensuring that in the limit @xmath13 the ground state is metallic . at the mean - field level both routes to obtain a metallic state ( doping and above describe depletion scheme ) lead to the same metallic ground state : a fermi - liquid with arbitrarily large effective mass ( or arbitrarily low coherence temperature ) as the coupling @xmath6 is reduced . the mean - field solution has several pitfalls . for instance , the imaginary part of the self - energy vanishes at all temperatures , the kondo scale corresponds to a phase transition rather than to a crossover scale and charge fluctuations on impurity sites are allowed since constraints are imposed only on average . for depleted kondo lattices it is possible to carry out auxiliary field quantum monte carlo ( qmc ) simulations without encountering the infamous sign problem . this comes with the restriction of particle - hole symmetry for the conduction band . nevertheless this gives us the opportunity to compare _ exact _ results with mean - field approximations . there are two challenges to circumvent before achieving good results . the first one is the implementation of the exchange interaction @xmath0 along with the constraint of singly occupied impurity sites . this problem has been solved in @xcite and extensively discussed in @xcite . the second challenge lies in controlling finite size effects stemming from the conduction electrons . this is extremely important in the framework of kondo physics since screening takes place on an energy scale set by the kondo temperature @xmath14 which is exponentially small in the weak coupling limit . if the level spacing of the conduction electrons exceeds this small energy scale , finite size effect will dominate the result . in section [ sizeb.chap ] we present a simple method to reduce by up to an order magnitude in temperature the onset of size effects in thermodynamic quantities ( specific heat , spin and charge susceptibilities ) for a given lattice size . section [ qmc_res ] summarizes our numerical results . ground state , thermodynamic and transport properties are computed on lattice sizes up to 500 sites . we show that our simulations reproduce the typical form of the resistivity versus temperature curve observed in heavy fermion materials : an initial rise as a function of decreasing temperature originating from the onset of screening of the magnetic impurities then followed by a sharp downturn signaling the onset of coherence between individual screening clouds . energy scales as a function of coupling strength are computed . the last section is devoted to conclusions and discussion of our results . our starting point is the two - dimensional kondo lattice model ( klm ) : @xmath15 here @xmath16 runs over the lattice sites of a two dimensional square lattice : @xmath17 , @xmath18 with @xmath19 the lattice vectors . the set of lattice points @xmath20 denote the positions of impurity spins . @xmath21 creates a conduction electron on site @xmath22 with z - component of spin @xmath23 and @xmath24 are the spin 1/2 operators with @xmath25 the pauli spin matrices . a similar definition holds for @xmath26 . since the klm stems from the strong coupling limit of the periodic anderson model @xcite the charge degrees of freedom on the f - sites are frozen . hence , a constraint of one electron per @xmath27-site holds for klm . in the notation of eq . [ klm_ham ] the usual klm model is obtained when @xmath28 ( see fig . [ latt.fig]a ) . in principle , the depletion scheme of impurity spins can be random - a situation which would reflect the experimental realization - or regular . as a first step we have chosen a regular depletion and considered @xmath29 + m \left [ \vec{a}_x + 2 \vec{a}_y \right ] $ ] ( see fig . [ latt.fig]b ) . from now onwards we will refer to this model as the depleted kondo lattice model ( dklm ) . the unit cell of the dklm has five orbitals : one localized for the impurity spin and four delocalized for the conduction electrons . the bravais lattice is spanned by the lattice vectors : @xmath30 and @xmath31 . , @xmath32 . ( b ) the depleted klm ( dklm ) . the corresponding bravais lattice is spanned by @xmath33 and @xmath34 . each unit cell contains a single localized orbital for the impurity spin and four delocalized orbitals accommodating the conduction electrons . in both ( a ) and ( b ) the top layer denotes the conduction electrons with hopping matrix element @xmath1 . the bottom layer are the magnetic impurities which couple to the conduction electrons via the exchange @xmath0 . ( c ) the brillouin zone of the dklm . the reciprocal lattice vectors read @xmath35 and @xmath36 , scaledwidth=45.0% ] having defined the model , we now address the question : to which degree does the klm at @xmath37 have the same physics as the dklm at @xmath38 ? here @xmath39 corresponds to the number of conduction electrons per unit cell . as mentioned previously , in the strong coupling limit the klm with band filling @xmath39 maps onto the @xmath40 hubbard model with band filling @xmath41 @xcite . the nature of the metallic state is not clear due to the strong correlations . however for fillings far from the doping induced mott metal - insulator transition , @xmath42 , it is reasonable to believe that beyond the one - dimensional case the ground state is a fermi liquid as described for example by a gutzwiller approximation . due to the special choice of the depletion , the strong coupling dklm at @xmath38 is a fermi liquid . in this limit , each impurity spin will capture a conduction electron and the remnant conduction electrons will form the metallic state . to be more precise at @xmath43 we obtain a three band model , where a single conduction band crosses the fermi level , @xmath44 . the thus obtained strong coupling fermi line is shown in fig . [ latt.fig]c . in the weak coupling limit , we can address the above question within a mean - field approximation @xcite . one can write the klm and dklm models as : @xmath45 where the last term accounts precisely for the spin flip terms : @xmath46 . with the above rewriting of the klm we can introduce order parameters both for magnetism and kondo screening . @xmath47 where @xmath48 correspond to the relevant wave vector for magnetic order . concentrating in a first step on paramagnetic solutions @xmath49 we obtain the mean - field hamiltonian @xmath50 where @xmath51 corresponds to the chemical potential , @xmath52 to a lagrange multiplier enforcing the constraint of single occupancy on the @xmath27-sites on average and @xmath53 is the number of impurity spins . the saddle point equations , @xmath54 where @xmath55 is the free energy may then be solved to obtain the mean - field solution . the thus obtained saddle point corresponds precisely to that obtained in the large - n approximation @xcite . ( a ) the band structure . energies are measured with respect to the fermi energy . the inset corresponds to the quasiparticle weight of the conduction band . ( b ) fermi line . ( c ) optical conductivity . note that there is a drude peak at @xmath56 . ( d ) drude weight normalized by the sum rule as well as the coherence temperature as a function of @xmath6 . the inset corresponds to the specific heat from which we obtain the coherence temperature . ( see text ) , title="fig:",scaledwidth=23.0% ] ( a ) the band structure . energies are measured with respect to the fermi energy . the inset corresponds to the quasiparticle weight of the conduction band . ( b ) fermi line . ( c ) optical conductivity . note that there is a drude peak at @xmath56 . ( d ) drude weight normalized by the sum rule as well as the coherence temperature as a function of @xmath6 . the inset corresponds to the specific heat from which we obtain the coherence temperature . ( see text ) , title="fig:",scaledwidth=23.0% ] + ( a ) the band structure . energies are measured with respect to the fermi energy . the inset corresponds to the quasiparticle weight of the conduction band . ( b ) fermi line . ( c ) optical conductivity . note that there is a drude peak at @xmath56 . ( d ) drude weight normalized by the sum rule as well as the coherence temperature as a function of @xmath6 . the inset corresponds to the specific heat from which we obtain the coherence temperature . ( see text ) , title="fig:",scaledwidth=23.0% ] fig . [ klm_pm ] summarizes the paramagnetic mean - field solution for the klm at @xmath57 . the band structure plotted in fig . [ klm_pm]a consists of two bands ( hybridized bands ) . as apparent , the band at the fermi energy is flat and has a small quasiparticle weight ( fig . [ klm_pm]a inset ) . the corresponding fermi surface ( fig . [ klm_pm]b ) , covers @xmath58 of the brillouin zone , so that both @xmath59electrons and conduction electrons are included in the luttinger volume . given the band structure , the optical conductivity follows ( fig . [ klm_pm]c ) . the large mass ( i.e. combined small quasiparticle weight and flat dispersion relation ) of the conduction electrons at the fermi energy leads to a small drude weight . the feature at higher frequencies are nothing but interband transitions . the drude weight normalized by the sum rule as a function of @xmath6 is plotted in fig . [ klm_pm]d . at small values of @xmath6 it becomes arbitrarily small reflecting the large mass of the charge carries . along with the small drude weight one expects a small coherence temperature . the coherence temperature ( @xmath60 ) - the energy scale at which the fermi liquid character of the ground state becomes apparent - may be defined for example by the energy scale below which the specific heat is linear in @xmath61 ( inset fig . [ klm_pm]d ) . as expected the @xmath6 dependence of @xmath62 tracks the drude weight . in the mean - field approach the kondo temperature , @xmath63 , corresponds to the temperature scale at which the order parameter @xmath64 vanishes . below @xmath63 the magnetic impurities are screened . in the mean - field approximation , this screening does not take place via the formation of a many - body state of the impurity spin and conduction electrons but via charge fluctuations on the impurity sites . in this approximation , charge fluctuations are allowed since the constraint of single occupancy of @xmath27-sites is only imposed on average . as apparent by the jump in the specific heat ( see inset of fig . [ klm_pm]d ) @xmath63 corresponds to a a phase transition rather than to a crossover scale . for a recent discussion of the kondo and coherence scales within this approximation , the reader is referred to @xcite . . ( i.e. the unit cell contains four conduction electrons ) ( a ) optical conductivity . ( b ) drude weight normalized by the sum rule . ( c ) specific heat versus @xmath65 . , title="fig:",scaledwidth=22.0% ] . ( i.e. the unit cell contains four conduction electrons ) ( a ) optical conductivity . ( b ) drude weight normalized by the sum rule . ( c ) specific heat versus @xmath65 . , title="fig:",scaledwidth=22.0% ] + . ( i.e. the unit cell contains four conduction electrons ) ( a ) optical conductivity . ( b ) drude weight normalized by the sum rule . ( c ) specific heat versus @xmath65 . , title="fig:",scaledwidth=22.0% ] we now turn our attention to the dklm . [ dklm_pm]a , b plot the optical conductivity as well as the drude weight versus @xmath6 within the paramagnetic mean - field approximation . as apparent , there is a clear similarity with the above presented results for the klm . from the specific heat data , ( fig . [ dklm_pm ] c ) one can extract the coherence temperature . here , some care has to be taken since the model has a van - hove singularity at the fermi level . hence the coherence temperature corresponds to the energy scale below which the specific heat follows a @xmath66 law . hence @xmath67 for the considered couplings . next , we consider magnetic degrees of freedom for the dklm . as mentioned above , at @xmath68 the model maps onto free electrons . the underlying particle - hole symmetry of the model directly leads to nesting so that the @xmath68 point will be unstable to magnetic ordering with wave vector @xmath69 ( fig . [ latt.fig]c ) . at weak couplings the ruderman - kittel - kasuya - yossida ( rkky ) interaction favors magnetic order rather that kondo screening @xcite . hence , for all values of @xmath6 we anticipate a magnetically ordered ground state . [ scales_mf.fig ] plots the mean - field transition temperatures corresponding to kondo screening , @xmath63 , ( below @xmath63 , @xmath70 ) and magnetism , @xmath71 ( @xmath72 @xmath73 ) . as apparent , magnetic order is always present . for values of @xmath74 the mean - field kondo scale exceeds the magnetic scale @xmath71 . for this parameter range and as a function of decreasing temperature , one expects to be able to study the formation of the heavy electron state prior to the onset of magnetic fluctuations which due to nesting will drive the system to an insulator . the qmc simulations presented in section [ qmc_res ] are aimed at studying this crossover . . @xmath63 corresponds to the kondo scale below which @xmath70 . @xmath71 corresponds to a magnetic energy scale below which magnetic ordering with wave vector @xmath75 schematically shown in fig . ( b ) occurs . @xmath63 and @xmath71 mark phase transitions . for @xmath74 both kondo screening ( @xmath76 ) and magnetic ordering ( @xmath77 , @xmath73 ) are present in the ground state . at @xmath78 the kondo scale vanishes since in the mean - field approximation magnetic ordering freezes the impurity spins . the coherence temperature ( not plotted in the figure ) lies roughly an order of magnitude lower that the kondo scale . hence for most of the considered couplings it will be masked by the magnetic instability . , scaledwidth=45.0% ] in this section we will first briefly summarize the numerical techniques used . we then turn our attention on a simple method to dramatically reduce the size effects for the simple case of free electrons . this is the key for producing size independent results down to low temperatures for the dklm . however , this method is general and may prove useful for other applications . we have used ground state and finite temperature versions of the auxiliary field qmc method to investigate the dklm . the @xmath79 method is based on the equation : @xmath80 where @xmath81 is required to be non - orthogonal to the ground state @xmath82 . finite temperature properties are calculated in the grand canonical ensemble via : @xmath83 } { { \rm tr } \left [ e^{-\beta h_{dpklm } } \right ] } .\ ] ] details of how to implement the qmc method without generating a sign problem in the particle - hole filled case have been introduced an extensively described for the klm in refs @xcite . since precisely the same _ tricks _ may be used for the dklm , we refer the reader to those articles for further details . we now concentrate on another important issue , size effects , which become extremely severe when one wishes to study metallic states down to low temperatures . finally , to determine the single impurity kondo scale for our given band - structure , we have used the hirsch - fye impurity algorithm @xcite . the same formulation as the lattice problem is readily implemented in this approach . reducing size effects in numerical simulations is important , since many models of correlated electron systems may be solved numerically only on _ small _ lattices . size effects become particularly severe when the ground state turns out be be a metallic state with large coherence temperature . on the other hand , insulators are characterized by the localization of the wave function and are hence rather insensitive to boundary conditions on finite sized systems . it thus becomes apparent , that the worst case scenario for severe size effects are just free electrons in a tight binding approximation : @xmath84 in many cases before turning on the interaction which will automatically restrict the size of the lattice under consideration it is important to control size effects for this simple case . we will concentrate on the two dimensional case on a torus geometry which for the above model reduces to imposing periodic boundary conditions : @xmath85 , @xmath86 where @xmath87 is the linear length of the lattice lying in the @xmath88 plane . to reduce size effects on thermodynamic quantities one may in principle consider the hamiltonian : @xmath89 where @xmath90 are arbitrary hopping parameters which have to satisfy @xmath91 clearly this choice of hopping matrix elements on finite lattices will break the lattice symmetry . this is a price we are willing to pay provided that the convergence as a function of system size of thermodynamics quantities is greatly improved . [ hopl ] nevertheless guarantees that in the thermodynamic limit this symmetry is restored . to determine the hopping matrix elements @xmath92 so as to reduce size effects on say the specific heat , @xmath93 , one may minimize @xmath94 ^ 2\ ] ] where the sum extends over a given range of temperatures . taking into account only amplitude modulations of the hopping matrix elements leads already to a cumbersome minimization problem which does not provide satisfactory results . instead of carrying out a complicated minimization problem we can try to guess which matrix elements @xmath90 will minimize size effects . it turns out that introducing a magnetic field produces remarkable results . the magnetic field is introduced via the peirls phase factors : @xmath95 with @xmath96 and @xmath97 the flux quanta . the torus geometry imposes restrictions on the @xmath98 field . since , a translation in the argument of the vector potential may be absorbed in a gauge transformation : @xmath99 we chose , the boundary condition @xmath100 to satisfy the requirement : @xmath101 = [ h(l),t_{l \vec{e}_y } ] = 0.\ ] ] here , @xmath102 corresponds to a translation by @xmath103 . however , magnetic translation operators belong to the magnetic algebra @xcite : @xmath104 thus , to obtain a single valued wave function the condition of flux quantization has to be satisfied : @xmath105 where @xmath106 is an integer . here , we consider a static magnetic field running along the @xmath107-axis perpendicular to the @xmath108 plane in which the lattice lies . hence , the smallest magnetic field which we can consider on a given lattice size satisfies : @xmath109 with this choice of magnetic field and associated vector potential eq . [ hopl ] holds . with ( right column ) and without ( left column ) magnetic field . here , we consider @xmath110 . , scaledwidth=45.0% ] to illustrate the reduction of size effects caused by the inclusion of the magnetic field , we first consider the single particle density of states . in a basis where @xmath111 is diagonal , @xmath112 with @xmath113 and @xmath114 , the local density of states reads : @xmath115 where @xmath116 is a positive infinitesimal and @xmath117 the total number of sites . since the magnetic field breaks translation invariance ( it is the site dependent vector potential which enters the hamiltonian ) @xmath118 is site dependent . averaging over sites yields the density of states @xmath119 plotted in fig . [ nom.fig ] . as apparent , without the magnetic field and up to @xmath120 , @xmath119 is dominated by size effects for the considered value of @xmath110 . in contrast , the presence of the magnetic field provides remarkable improvements . in particular the van - hove singularity is well reproduced already on @xmath121 lattices and at @xmath120 the result is next to exact for the considered value of @xmath122 . it is instructive to look at the @xmath123 case with and without magnetic fields . when @xmath124 is turned on , the degeneracy of levels is lifted . each level - apart from the @xmath125 level which is two fold degenerate - is nondegenerate . this is precisely what one expects for landau levels which have degeneracy @xmath126 which is unity in our case . this provides an intuitive understanding of why the method works so well . since each level becomes singly degenerate , the single particle states cover homogeneously the the energy range of the band - width . clearly this can only be achieved by breaking the lattice symmetry on finite sized systems . we now turn our attention to the specific heat coefficient @xmath127 ( see fig . [ cv.fig]a,b ) . as apparent , for a given system size , the inclusion of the magnetic field provides a gain of more than one order of magnitude in the temperature scale at which size effects set in . in particular the @xmath128 behavior of @xmath129 due the the van - hove singularity becomes apparent already on @xmath130 lattices . + upon inspection a similar improvement is obtained for the spin susceptibility ( see fig . [ cv.fig]c,d ) . note that since we are dealing with free electrons the charge and spin susceptibilities are identical . one crucial question is whether the magnetic field will introduce a sign problem in the numerical simulations . it turns out that for dklm at the particle - hole symmetric point it does not @xcite . hence we can incorporate it in our simulations and benefit from the above demonstrated drastic improvement in reduction of size effects . other schemes have been proposed to reduce size effects . in particular , averaging over boundary conditions has been suggested @xcite . this method has the advantage of not breaking translation symmetry . however , the averaging requires several simulations and is hence computationally expensive . in contrast with the presented method the improvement in reduction of size effects is obtained within a single simulation . our numerical results for the dklm are summarized in this section . we first concentrate on ground - state properties and check for magnetic ordering as a function of @xmath6 . various thermodynamic quantities ( specific heat , spin and charge susceptibilities ) as well as the optical conductivity and hence resistivity are then analyzed as a function of temperature at different values of @xmath6 . to detect long - range magnetic ordering , we compute the spin - spin correlation function at @xmath79 : @xmath131 here we consider @xmath75 and the sum runs over the impurity spins . in the presence of long - range magnetic order , @xmath132 is an extensive quantity so that the staggered moment , @xmath133 , is finite . @xmath134 denotes the number of unit - cells . + at @xmath79 for the dklm as a function of system size and various values of @xmath6 . the @xmath135 point corresponds to a lattice of @xmath136 unit cells and hence 500 orbitals . after extrapolation to the thermodynamic limit , we obtain the staggered moment plotted in the inset . , scaledwidth=45.0% ] figure [ spint0.fig ] plots @xmath137 for various values of @xmath6 . the inset in fig . [ spint0.fig ] plots the staggered moment obtained after extrapolation of @xmath137 to the thermodynamic limit . the data is thus consistent with long - range magnetic ordering at least up to @xmath138 . clearly as @xmath6 grows and the staggered moment becomes small , it is increasingly hard to distinguish it from zero . nevertheless , and based on the mean - field result , we expect the staggered moment to vanish only in the limit @xmath13 . following arguments put forward in ref . @xcite kondo screening and magnetic ordering to coexist : impurities are partially screened and the remnant magnetic moment orders . depending upon the relative energy gains of both mechanisms a staggered moment of arbitrary magnitude can be produced . in the weak coupling limit , the energy gain for magnetism is the rkky interaction . this interaction leads to an effective heisenberg model between impurity spins with exchange @xmath139 where @xmath140 corresponds to the spin susceptibility of the conduction electrons . on the other hand screening occurs on an energy scale set by the kondo temperature : @xmath141 where @xmath142 is the band width . hence for small values of @xmath6 the system will gain energy by ordering magnetically . although the wave vector of magnetic ordering will depend on the details of the spin susceptibility , this source of energy gain is present irrespective of the band structure . on the other hand , the strong coupling result is an artifact of the model . in this limit , if kondo screening were complete , a nested fermi surface would be produced as in the @xmath13 limit . hence , in the ground state , the system will always gain energy by keeping kondo screening partial . the remnant magnetic moment will order with the nesting wave vector thereby opening a gap and gaining energy . were it not for nesting a finite critical value of the coupling @xmath6 at which a quantum phase transition between ordered and disordered magnetic states would occur . unfortunately , this is beyond the scope of the qmc since the absence of particle - hole symmetry and hence nesting leads to a sign problem . here , we consider finite temperature properties of the model . as discussed above , the ground state is a magnetic insulator irrespective of @xmath6 . the salient features of the model however , lie in its thermodynamic properties . we first consider the real part of the optical conductivity as obtained from the kubo formula , @xmath143 . this quantity is related to the imaginary time current - current correlation functions via : @xmath144 here @xmath0 is the current operator along the @xmath12 or @xmath145 lattice direction and @xmath146 represents an average over the finite - temperature ensemble . the above inverse laplace transform , to obtain the optical conductivity is carried out with the me @xcite method . we use the paramagnetic mean - field solution with substantial broadening as default model . note that the order parameter @xmath147 which is proportional to the spin - spin correlation between the conduction electron and impurity spin is estimated from the monte carlo run and not by solving the saddle point equations . . the data in figs . ( b),(c ) stem from simulations on lattices with @xmath148 unit cells ( l=8 ) . in fig . ( a ) both @xmath130 and @xmath149 results are included . as apparent and due to the inclusion of the magnetic field ( see section [ sizeb.chap ] ) size effects are next to absent until the magnetic length scale exceeds the lattice size ( see text ) . , scaledwidth=45.0% ] our results for the real part of the optical conductivity and associated resistivity @xmath150 are plotted in fig . [ cond16yy.fig ] . here we consider @xmath151 . at high temperatures , @xmath152 , the data consists of a single drude feature . as the temperature is lowered to @xmath153 , the weight under the drude peak is reduced , and a feature at @xmath154 is formed . a rise in the associated resistivity as a function of decreasing temperature is observed in this temperature range . this rise in resistivity originates from singular spin flip scattering on impurity spins and is reminiscent of the single impurity physics . by further reducing the temperature to @xmath155 the drude feature becomes sharper thus leading to a drop in the resistivity . finally , at our lowest temperatures ( @xmath156 ) there is a marked upturn in the resistivity . at those temperatures the optical weight under the drude peak starts to shift to frequencies @xmath157 . . , scaledwidth=45.0% ] the overall form of the resistivity versus temperature curve ( fig . [ cond16yy.fig]a ) allows us to define three temperature scales . the highest energy scale we will discuss corresponds to the resistivity minimum , @xmath158 . for the considered coupling @xmath159 . we will define @xmath160 as the temperature scale at which a maximum in the resistivity is observed . the magnetic scale @xmath161 is defined by the energy scale below which the @xmath162 follows an _ activated _ behavior . next , we consider thermodynamic properties and concentrate on their behavior at the above defined energy scales . [ ther16.fig ] plots the specific heat @xmath163 , charge and spin uniform susceptibilities ( @xmath164 , @xmath165 ) as well as the staggered spin susceptibility ( @xmath166 , @xmath75 ) as a function of temperature at @xmath151 . from the technical point of view , the specific heat is computed using the recently proposed me based method of ref . @xcite . as apparent in the temperature region @xmath167 one observes the following features . i ) as indicated in fig . [ ther16.fig]c the specific heat is consistent with a @xmath168 law . ii ) the uniform spin and charge susceptibilities after going through a broad maximum seem to saturate . both points are the characteristics of a fermi liquid . the staggered spin susceptibility essentially measures the magnetic length scale at the considered wave vector . the inset of fig . [ ther16.fig]a shows a marked increase in this quantity at the spin scale @xmath71 . at the same energy scale , sharp _ peak in the spin susceptibility is observed . following the size effects , the data is consistent with a saturation of this quantity in the limit of zero temperature . at @xmath71 the charge susceptibility [ ther16.fig]b decreases rapidly . upon analysis of size effects the data is consistent with the vanishing of this quantity at @xmath79 . upon close analysis , an anomaly in the specific heat is apparent at @xmath71 . thus , @xmath71 marks the onset of substantial spin fluctuations which triggers the formation of an insulating state . that is the vanishing of the charge susceptibility and an activated behavior of the resistivity below @xmath71 . . the inset plots the resistivity versus temperature . for those simulations the lattice size is @xmath130 . , scaledwidth=45.0% ] but for @xmath169.,scaledwidth=45.0% ] next , we consider smaller couplings , @xmath169 . transport and thermodynamic properties are plotted in figs . [ cond08yy.fig ] and [ ther08.fig ] . as apparent from the conductivity data ( fig . [ cond08yy.fig ] ) , no maximum as a function of temperature is observed . after hitting a minimum at @xmath170 the resistivity grows monotonically as a function of decreasing temperature . the thermodynamic data of fig . [ ther08.fig ] equally shows no sign of the @xmath2 scale . that is , there is no energy scale for which the specific heat has a linear in @xmath61 dependence . the only scales which are apparent in figs.[cond08yy.fig ] and [ ther08.fig ] are resistivity minimum @xmath158 and the spin scale @xmath71 . at @xmath71 an anomaly in @xmath163 is detected , a drop in @xmath171 is observed and @xmath165 exhibits a sharp peak . the peak in @xmath165 is again located at the temperature scale at which @xmath166 shows a sharp increase . our qmc results are summarized in fig . [ scales_qmc.fig ] . the highest energy scale , @xmath158 , corresponds to the resistivity minimum , triggered by enhanced spin flip scattering off the impurity spins . as already observed for the klm , this energy scale in a first approximation tracks @xmath0 . at high temperatures , it is tempting to try to understand this energy scale in terms of the single impurity kondo model . in this case , we can use the closed form for the impurity resistivity ( see @xcite and references theirin ) : @xmath172^{1/2 } } \right]\ ] ] where @xmath63 corresponds to the kondo scale , @xmath173 and @xmath174 is the bandwidth . assuming a power - law , @xmath175 , for the resistivity of the conduction electrons , the resistivity minimum will scale as @xmath176 at weak couplings . given the above , our qmc result @xmath177 follows by setting @xmath178 and we are left with searching for mechanisms which will lead to a linear in @xmath61 resistivity . a possibility is scattering on spin fluctuations @xcite . note that both the klm @xcite and dklm an enhancement of local antiferromagnetic spin - spin correlations between impurity spin and conduction electron ( @xmath179 ) is observed at an energy scale tracking @xmath158 . hence , magnetic fluctuations are potentially present at this energy scale . clearly further work is required to justify this hand waving argument . is enhanced in the dklm . depleting further the dklm of impurity spins , we should ultimately arrive to the single impurity case were @xmath158 scales to infinity when the conduction band is modeled by free electrons . ] . @xmath158 denotes the resistivity minimum . @xmath180 corresponds to a crossover scale at which a maximum in resistivity is observed . below @xmath71 at marked increase in the magnetic length scale is observed . we use the form @xmath181 to fit the @xmath182 and the form @xmath183 to fit the data for @xmath158 . the crosses correspond to the kondo temperature of the single impurity problem . ( see section [ tki ] ) , scaledwidth=40.0% ] below @xmath158 the resistivity grows as a function of decreasing temperature . for values of @xmath74 , the scale @xmath2 emerges . in particular and for the considered couplings , below @xmath2 the resistivity drops sharply , the specific shows a linear in @xmath61 behavior , and both the spin and charge susceptibilities seem to converge to finite values . in a first approximation , this is what expects for a fermi liquid . given the limited data , it is hard to pin down the functional form of @xmath2 as a function of @xmath6 . nevertheless , an exponential fit ( @xmath184 ) accounts reasonably well for the data . due to the particle hole symmetry of the model which automatically leads to nesting , we can not follow the metallic state to @xmath79 . at @xmath71 substantial spin - spin fluctuations set in and drive the system to an insulator with activated resistivity . for values of @xmath78 , @xmath2 drops below @xmath71 . hence after a slow increase of resistivity below @xmath158 , an activated behavior is expected below @xmath71 . it is now interesting to compare @xmath2 with the kondo temperature of the single impurity problem : @xmath185 to obtain the kondo temperature one can compute the impurity spin susceptibility : @xmath186 where @xmath187 . in the low temperature limit , @xmath188 is the only scale hence @xmath189 where @xmath55 is a universal function . lattice . , scaledwidth=40.0% ] in fig . [ tk_imp.fig ] and for the three considered values of @xmath6 a good collapse of the data is obtained with the listed values of the kondo temperature @xmath190 . comparison with fig . [ scales_mf.fig ] shows that the mean - field approximation produces next to identical results for the kondo temperature . recall that the kondo temperature for the impurity and lattice cases are identical in the mean - field approach . the important point however , is that within our accuracy @xmath2 compares very well with @xmath188 . this stands in rough agreement with experiments on ce@xmath9 la@xmath191 cu@xmath11 @xcite . here the single impurity kondo temperature ( @xmath192 ) is given by @xmath193 . as @xmath12 grows ( @xmath194 ) , a maximum in the resistivity versus temperature curve develops at energy scales in the range @xmath195 . it is also worth pointing out that below @xmath2 the spin susceptibility in cecu@xmath11 seems to saturate @xcite . this stands in rough agreement with our results . hence , @xmath2 corresponds essentially to the single impurity kondo temperature @xmath188 below which the magnetic impurities are screened . it marks the onset on the heavy - electron state . to summarize , we have presented a detailed numerical study of a depleted kondo lattice . the depletion considered here is very special , and corresponds , to the best of our knowledge , to the minimal regular depletion required to obtain a two - dimensional metallic state in the @xmath43 limit . from the technical point of view , the qmc approach is flexible and can be applied to any depletion pattern , random or regular . the constraint to avoid the minus sign problem is the particle - hole symmetry of the conduction band . hence various subjects may be studied . as function of depletion concentration , a metal - insulator transition should occur as observed in ce@xmath4bi@xmath5pt@xmath4 when ce is replaced by la @xcite . in the strong coupling limit one expects this metal - insulator transition to occur at the percolation threshold . furthermore , in the insulating phase random depletion of the @xmath27-sites should trigger magnetic ordering between the magnetic moments induced by the depletion . hence order by disorder is equally a phenomena which can studied in this model . to control size effects , we have introduced a magnetic field perpendicular to the two dimensional lattice . we have shown that by scaling the magnetic field as @xmath196 extremely good size scaling to the thermodynamic limit is obtained . in particular already on small lattices the temperature scale at which size effects set in drops by an order of magnitude when the magnetic field is included . this is the key to our results since it opens the energy window at which coherence effects may be studied . the qmc results for the dklm show three temperature scales which are summarized in fig . [ scales_qmc.fig ] . i ) the largest scale , @xmath158 , corresponds to the minimum in the resistivity and tracks @xmath6 . ii ) @xmath71 corresponds to the energy scale below which the magnetic length scales grows presumably exponentially with decreasing temperature . due to particle - hole symmetry and associated nesting properties those spin fluctuations drive the system to an insulating state . the ground state of the model remains a magnetically ordered insulator irrespective of the value of @xmath6 . as in the klm @xcite our data supports @xmath197 at weak couplings , in accordance with the rkky interaction . in the strong coupling limit @xmath71 is exponentially suppressed and vanishes at @xmath68 . note that in the absence of nesting one would expect the occurrence of a quantum phase transition between magnetically ordered and paramagnetic metallic states at a finite values of @xmath6 . this is unfortunately not accessible to the qmc method due to severe sign problems . iii ) the temperature scale @xmath2 as defined by the maximum in the resistivity versus temperature cure becomes apparent when it exceeds the spin scale @xmath198 . this happens at @xmath74 . in the temperature range @xmath199 we see signatures of fermi liquid character : the specific heat is roughly linear in @xmath61 and the spin and charge susceptibilities appear to converge to finite values . unfortunately , we can not follow this metallic phase down to @xmath79 because of the onset of spin fluctuations below @xmath71 . iv ) we have computed the corresponding single impurity kondo temperature ( @xmath188 ) with the hirsch - fye qmc algorithm . within the considered small intermediate coupling range , our results suggest @xmath200 . comparison with the mean - field results ( see fig . [ scales_mf.fig ] ) shows that the spin scale is well reproduced . clearly , the @xmath158 scale is absent at the mean - field level . the mean - field kondo temperature @xmath63 is independent on the impurity density and hence is identical to that of the single impurity model . comparison with _ exact _ qmc results shows that this mean - field approximation yields next to exact values of the single impurity kondo temperature . hence the mean - field kondo temperature which marks a phase transition in this approximation evolves to the crossover scale @xmath2 when the model is solved exactly . finally we comment on the coherence scale . at first glance it is tempting to associate the coherence scale with @xmath2 since signs of fermi liquid character start emerging at this energy scale . however this identification will lead to contradictions . in particular , in the strong coupling limit the coherence scale scales to a finite valuec . the coherence temperature is then given by the coherence temperature of the free electron model which is finite . ] but @xmath2 is expected to track @xmath201 . hence we can only speculate that the coherence scale as in the mean - field approach lies well below @xmath2 and is masked by @xmath71 . we thank hlrs ( stuttgart ) for generous allocation of cpu time on the cray - t3e . i acknowledge useful discussions with s. capponi , m. feldbacher and a. muramatsu . the deutsche forschungsgemeinschaft ( dfg ) is thanked for financial support under the grant number as 120/1 - 1 .	we consider a two dimensional kondo lattice model with exchange @xmath0 and hopping @xmath1 in which three out of four impurity spins are removed in a regular way . at the particle - hole symmetric point the model may be studied with auxiliary field quantum monte carlo methods without sign problems . to achieve the relevant energy scales on finite clusters , we introduce a simple method to reduce size effects by up to an order of magnitude in temperature . in this model , a metallic phase survives up to arbitrarily low temperatures before being disrupted by magnetic fluctuations which open a gap in the charge sector . we study the formation of the heavy - electron state with emphasis on a crossover scale @xmath2 defined by the maximum in the resistivity versus temperature curve . the behavior of thermodynamic properties such as specific heat as well as spin and charge uniform susceptibilities are studied as the temperature varies in a wide range across @xmath2 . within our accuracy @xmath3 compares well to the kondo scale of the related single impurity problem . finally our qmc resuls are compared with mean - field approximations . + pacs numbers : 71.27.+a , 71.10.-w , 71.10.fd + +
the cessation of eclipses in a binary star is a rare phenomenon that can most often be explained by the presence of a third object in the system inducing perturbations in the orbital elements of the inner pair . the modulation of the inclination angle to the line of sight , in particular , is the direct cause of a change in the orientation of the plane of the binary such that eventually the stars no longer block each other s light at conjunction . few examples of this phenomenon are known , among them ay mus ( sderhjelm 1974 ) and v907 sco ( lacy , helt & vaz 1999 ) . an especially interesting case is ss lacertae ( hip 108981 , bd@xmath33782 , @xmath4 days , , @xmath5 , @xmath6 , epoch and equinox j2000 ) , which stopped eclipsing around the middle of the 20@xmath0 century . our knowledge about light variations in this binary when it _ was _ eclipsing comes entirely from visual and photographic measurements that go back more than a century , but that are unfortunately of rather poor quality . however , the object has received considerable attention recently as a result of a re - measurement of archival plate material and new spectroscopic observations . lehmann ( 1991 ) re - measured original plate material from the sonneberg observatory ( 1890 - 1989 ) , presenting the first evidence that the depth of the eclipses had changed over the years . he correctly interpreted this as due to the presence of an unseen third star in the system , which is gradually changing the inclination angle of the inner pair . tomasella & munari ( 1998 ; hereafter tm98 ) derived a double - lined spectroscopic orbit for the binary , confirming earlier reports that the system had not been disrupted by a chance encounter with another star and that the period had not changed , as had been proposed initially to explain the disappearance of eclipses . milone et al . ( 2000 ; hereafter m00 ) carried out an exhaustive analysis of all available light curves of ss lac , and presented clear confirmatory evidence of changes in the depth of the eclipses using their new measurements of the original harvard observatory patrol camera plates ( see dugan & wright 1935 ; hereafter dw35 ) as well as other published measurements . further extensive spectroscopic observations were presented by torres & stefanik ( 2000 ; hereafter ts00 ) , who showed that the system is indeed triple . they determined the orbital elements of the distant third object , with a period of about 679 days and a slightly eccentric orbit . they also detected apsidal motion for the first time , which is another manifestation of perturbations caused by the third star . further properties and the history of observations of the system are described in detail by m00 and ts00 . adding to its interest , ss lac is a member of the open cluster ngc 7209 , and therefore other information such as estimates of the age , distance , and metal abundance is available . conversely , accurate dimensions for the binary components , if they can be determined , offer a chance to perform critical tests of stellar evolution models as well as valuable clues on the cluster itself , since ss lac is located very close to the turnoff point in the h - r diagram . as pointed out by ts00 , their high - quality spectroscopic orbital solution for ss lac has the potential of giving very precise absolute mass determinations if the inclination angle at the current epoch can be established accurately enough . spectroscopic observations only allow the minimum mass of each star to be determined , in the form of the quantity @xmath7 . the value of the inclination at the present time can only be inferred by extrapolation , and this requires in particular that its _ rate of change _ be known very accurately . estimates have been made in a variety of ways by a number of authors , as described later , but are typically based on very limited data . the motivation for this study is precisely to use the new information that has now become available in the form of light curves analyses and measurements of the amplitude of the eclipses , to improve our knowledge of the time dependence of the inclination angle and thus to allow the absolute masses and also the radii of the component stars to be established . we explore ways of doing this and apply a physically realistic model based on the known properties of the binary and the effects of the third star expected from theory . although as we describe later this potential can still not be realized because of limitations in the observations , new insights can be gained on the timescale and the pattern of the appearance and disappearance of eclipses . m00 presented measurements of the depth , @xmath8 , of the eclipses in ss lac that show convincingly that the amplitude of the minima decreased gradually from the 1890 s and early 1900 s to shortly before the middle of the 20@xmath0 century , when eclipses were last recorded . the peak amplitude is believed to have occurred around 1910 or so . we will re - examine these data in detail in 3 , but we focus first on a procedure to extract information on the rate of change of the inclination angle from these measurements . as a means of estimating the approximate time of onset and cessation of eclipses , or the beginning and end of the _ eclipse season _ , as they termed it , m00 presented a quadratic fit to the measurements around the primary eclipse as a function of time ( their fig . 7 ) that provides a reasonable representation of the data . they correctly pointed out , however , that a quadratic fit is not necessarily a realistic model for the true variation , which is more likely to taper off at either side of the maximum rather than end abruptly . as an alternative , m00 discussed a fit to a model involving the complementary error function^2\right\}$ ] . the quadratic argument makes the function symmetrical around the maximum . for the benefit of other readers we note also that the phases listed in their table 10 and table 11 , while computed from the tm98 ephemeris , have an integer number of cycles added ( 2713 ) to render them positive . also , the cycle number for their primary minimum # 40 should read 2421.988 . ] , which has the desired property of decreasing slowly at both ends although it still lacks any physical basis and is merely a mathematical tool . the true variation of the depth of the eclipses may be separated into two parts : a function we call @xmath9 ( one for each minimum ) , which describes the change in the amplitude as a function of the inclination angle @xmath10 , and a second function @xmath11 that describes the time variation of @xmath10 . this latter function is what we seek to determine from the observations . we will assume here that the inclination angle is the only orbital element that changes significantly with time due to the third star , although strictly speaking this may not be true for ss lac . on the one hand apsidal motion has been detected ( ts00 ) , and the secular variation in @xmath12 ( the longitude of periastron ) will alter the shape of the light curve slightly over long timescales ( @xmath11000 yr ; see ts00 ) . on the other hand the perturbations from the third star can also cause a modulation of the eccentricity of the binary ( see , e.g. , mazeh & shaham 1979 ) , although this effect has not yet been detected . neither of these effects will change the amplitude of the eclipses appreciably in this case , so for our purposes all elements except @xmath10 may be assumed to be constant . similar arguments to describe the time variation of @xmath8 appear to have been used by lehmann ( 1991 ) in the first report presenting evidence of morphological changes in the light curve of ss lac , although few details are given . he correctly attributed the changes in the inclination angle to perturbations from a third object in the system , and estimated the timescale for the variation . the change in the depth of the eclipses as a function of the inclination angle can be derived in a fairly straightforward way once a fit to a representative set of observations has been made , using any of the current computer codes to analyze a binary light curve . one simply holds all elements fixed and changes @xmath10 to produce a family of synthetic light curves , and then measures the depth of the eclipses ( separately at both minima ) on each of these as a function of @xmath10 . a table can then be produced that represents @xmath9 , and this may be interpolated for ease of use . the critical step here is arriving at a light curve solution from the real observations that accurately represents the physical properties of the components . the very fact that the light curve is changing with time and the relatively poor quality of the historical brightness measurements of ss lac make this step more difficult than it would seem . light curve analyses of the photometric observations for ss lac have been described in detail by m00 . they used the university of calgary version of the wilson - devinney code ( wilson 1992 ; milone , stagg & kurucz 1992 ; stagg & milone 1993 ; kallrath et al . 1998 ) , and examined the three most complete data sets available : \1 . photographic measurements by dw35 based on harvard college observatory patrol camera plates , covering the interval 1890 - 1934 . these were published only in graphical form as an average light curve , already folded using the ephemeris by the authors . a sizeable fraction of these plates and subsequent plates taken at harvard were re - measured by schiller , bridges , & clifton ( 1996 ) , although these new measures were not used by m00 for their light curve analyses . photographic measurements by wachmann ( 1936 ) based on plates from bergedorf spanning the interval 1924 - 1932 . low light levels occur only during the last two years of this interval . visual estimates by kordylewski , pagaczewski , & szafraniec ( 1961 ) between 1927 and 1948 . in all cases the solutions were iterated with fits to the radial velocity measurements of tm98 , under various assumptions on the relative temperatures , luminosities , and mass ratio of the components . in their attempt to avoid smearing of the light curve due to real changes in shape as a function of time , m00 analyzed these three data sets separately . as a result some of them are rather sparse , which adds to the difficulties imposed by the poor quality of the observations . in addition , there are some inconsistencies between separate solutions , and in the end m00 adopted their results for the dw35 data set as being more reliable . a light curve solution for ss lac was also presented by ts00 , who chose to merge the datasets by dw35 and wachmann ( 1936 ) since they overlap in time . they used the computer program ebop ( etzel 1981 ; popper & etzel 1981 ) , which , although simpler than the wilson - devinney code , is quite adequate for this well - detached system . the m00 and ts00 solutions show some significant differences , no doubt a reflection of the difficulties mentioned above , but also because the m00 analyses were done without the benefit of the new spectroscopic information obtained by ts00 , which only appeared later . the light ratio ( secondary / primary ) derived by m00 is approximately 2.2 , while the spectroscopic measurements by ts00 indicate a value quite close to unity ( @xmath13 ) , with much larger ratios being clearly ruled out by those observations . tm98 seem to concur with ts00 , at least qualitatively , reporting that the secondary is slightly " more luminous than the primary from the relative depths of the doppler cores and h@xmath14 line wings in their spectra . because of the inherent difficulty of determining the ratio of the radii , @xmath15 , from light curves such as that of ss lac in which the stars are nearly equal and the eclipses only partial , ts00 incorporated their spectroscopic light ratio as an external constraint on their light curve solution . in this way they derived a ratio of the radii of @xmath16 ( secondary / primary ) . the unconstrained solution by m00 , on the other hand , leads to a value of @xmath15 around 1.5 , similar to that obtained originally by dw35 also without the use of an external constraint . such a large value seems somewhat unusual for main - sequence stars with very similar masses ( @xmath17 ; ts00 ) . we note also that the maximum amplitude of the minima from the m00 solution , setting @xmath18 , is @xmath10.45 mag ( see fig . 1 ) , whereas their empirical fit to the measured amplitudes , mentioned earlier , suggests peak values considerably higher than this . several of the measurements of the depth of the eclipses presented by the same authors ( their table 10 and table 11 ) also reach larger values ( as large as 0.76 in one case ) between 1900 and 1910 . similar indications are seen in the work by lehmann ( 1991 ) . admittedly some of these measurements suggesting deeper minima have rather large errors due to the photographic nature of the material , and two of them by m00 at about the epoch of maximum amplitude actually have much lower values , as do two independent measurements by lehmann . further light may be shed on this issue by examining the individual brightness estimates for ss lac from the harvard plates as re - measured by schiller et al . ( 1996 ) ( see also m00 ) . these measurements were kindly provided to the author by e. f. milone . 2 shows a sequence of those measurements from 12 harvard plates taken on the night of 1902 may 30 ( jd @xmath19 ) during a primary eclipse . the brightness estimates have been referred , for convenience , to the mean light level of the system outside of eclipse . comparison with the same curves shown in fig . 1 gives the clear impression that the primary minimum was indeed deeper than 0.45 mag , and that the data are considerably better represented by the ts00 curve . the solution by ts00 for @xmath18 indicates depths of about 0.7 mag , which is close to what is expected for component stars of roughly equal brightness . the adoption of a representative light curve solution has a rather significant impact on the results described below . although the ts00 solution is certainly not exempt from criticism , based on the arguments presented above we have chosen for the present analysis to adopt that solution as a more consistent representation of the physical properties of the stars in ss lac , pending improvements that may become possible as we discuss later in 6 . using ebop as described earlier we have determined the functions @xmath20 and d@xmath21 for the depths of the primary and secondary eclipse , respectively , which do not differ greatly . they are represented graphically in fig . the simplest approximation , and the one usually adopted , is that of a uniform variation of the inclination angle with time . as a convenient expression we adopt @xmath22 where @xmath23 is the epoch at which eclipses are central ( @xmath24 ) . in 5 we describe a more realistic representation but we show also that , given the quality of the data at our disposal , the linear expression is a sufficiently accurate description in the time interval covered by the present data . for other purposes such as large extrapolations or determining the boundaries of the eclipse season of ss lac , a linear change in @xmath10 is inadequate ( see 6 ) . several estimates of the slope @xmath25 have been made in recent years based on a variety of assumptions : \1 . lehmann ( 1991 ) used measurements of the eclipse amplitudes based on photographic plates from the sonneberg observatory in a procedure similar to what we described above , and obtained @xmath26 deg yr@xmath27 . he also estimated the eclipses to have been central in @xmath28 . few of the details are given , however , and the original data are not available except in graphical form . tm98 assumed typical radii for the stars ( @xmath29 r@xmath30 ) based on an adopted spectral classification . from the linear separation between the components derived from their spectroscopic orbital solution they estimated the inclination angle at the end of the eclipsing season to be @xmath31 . the authors assumed that the eclipses were central in 1900 - 1915 , and also that they disappeared at about 1960 based on lehmann s ( 1991 ) work . from this they arrived at @xmath32 deg yr@xmath27 . extrapolating to the mean epoch of their radial velocity measurements they estimated @xmath33 . m00 estimated the inclination angle at the mean epoch of the velocity measurements by tm98 to be @xmath34 , through a procedure involving iterations between separate spectroscopic and light - curve solutions using the photometric data by dw35 . adopting an epoch of 1912 for central eclipses , they then obtained @xmath35 deg yr@xmath27 . ts00 combined the photographic measurements by dw35 and wachmann ( 1936 ) to obtain a light curve solution giving @xmath36 , which they assigned to a mean epoch of @xmath37 . based on this analysis and on their spectroscopic orbit , they estimated the minimum inclination angle for eclipses to occur to be @xmath38 , and adopted 1951 as the epoch of cessation of eclipses . from this they derived @xmath39 deg yr@xmath27 . although all these estimates of @xmath25 turn out to be rather similar , they are not independent and the assumptions made are quite varied and sometimes inconsistent , mostly due to limitations in the observational material . in the sections that follow we describe our own efforts based on observations that have not been used before for this purpose . measurements of the depth of the eclipses in ss lac were compiled by m00 from several sources : their own re - measurement of many of the original harvard plates used by dw35 ; photographic observations published by wachmann ( 1936 ) , nekrasova ( 1938 ) , and mossakovskaya ( 1993 ) ; visual determinations by kordylewski et al.(1961 ) ; and more recent photoelectric measurements obtained by them at the rothney astrophysical observatory and also by mossakovskaya ( 1993 ) , after the eclipses stopped . a total of 43 estimates of @xmath8 were collected for the primary minimum , and 27 for the secondary minimum , over a period of nearly 100 years . some of them are averages determined from several individual brightness measurements . in addition to these , lehmann ( 1991 ) reported estimates of the amplitude of the eclipses from sonneberg observatory plates , but they were published only in graphical form . further photographic observations of ss lac were made by tashpulatov ( 1965 ) from 1937 to 1955 . however , these data have been called into question by mossakovskaya ( 1993 ) , and a re - examination of some of the original plate material by other investigators has cast similar doubts ( see m00 ) . because of this , and because they contain relatively few measurements at low light levels , we have not considered the tashpulatov observations here . with regard to the other measurements , for the purpose of this paper we ignore the minor difference between the visual and photographic bands , following m00 . the crucial ingredient in the selection of brightness measurements for estimating the amplitude of the eclipses is the ephemeris used to establish which observations are near an eclipse . m00 adopted the ephemeris by tm98 , min i @xmath40 , which was the most recent at the time . a much more accurate determination is now available from ts00 , based on all published times of eclipse as well as their new radial - velocity measurements , and accounting also for light travel time ( due to the presence of the third star ) and apsidal motion . the sidereal period ( @xmath41 days ) is significantly different from the previous value , and one order of magnitude more precise . over the time interval in which the eclipse amplitude measurements of m00 are non - zero ( @xmath11890 - 1940 , or 50 yr ) , the difference in the periods alone accumulates to a phase shift of 0.02 , almost exactly half of the width of an eclipse . in extreme cases a measurement regarded by m00 to be near mid - eclipse may actually fall beyond the limits of eclipse , leading to an underestimate of depth of the corresponding minimum . this is especially true given that m00 adopted the conservative approach ( precisely to allow for possible errors in the tm98 ephemeris ) of including all brightness measurements they determined to be within @xmath42 in phase from the minima , which is actually twice the true half - width of the eclipses . an illustration of this effect can be seen in fig . 4 . in the top panel we show 17 visual brightness estimates obtained near primary eclipse by kordylewski et al . ( 1961 ) on the night of 1930 november 28 ( filled symbols ) , phased with the ephemeris by tm98 . this is the longest published sequence of measurements made on a single night during an eclipse that shows a clear change in brightness . the light curve displayed for comparison purposes is from the solution adopted in 2.1 , with a maximum depth appropriate for the date of these observations . also shown with a different symbol are all measurements by wachmann ( 1936 ) that are near primary eclipse . these observations are confined to a 2-yr interval ( mean epoch @xmath11931.8 ) , as mentioned in 2.1 , that happens to be very close to the kordylewski date . the phase shift is fairly obvious for both data sets . when the same measurements are folded with the ephemeris by ts00 , on the other hand , the fit is considerably better ( lower panel ) . we point out , incidentally , that both the kordylewski observations and the wachmann observations are completely independent of the ephemeris by ts00 since they were not used by the authors in its derivation , and thus they serve as an independent check on its accuracy , @xmath11902.4 ) were folded with the tm98 ephemeris , yet they show good phase agreement with the predicted light curves , seemingly conflicting with the evidence above . as it turns out , if the phasing is done instead with the ts00 ephemeris the result is essentially the same because both prescriptions made use of the main epoch of primary minimum from dw35 in their determination , which happens to be the very night of these measurements . in the case of tm98 this epoch is imposed exactly , and for ts00 it enters more weakly as part of a least squares adjustment . the main difference will then be simply a phase shift , equal to the residual of that epoch of primary minimum from the ts00 fit ( see their table 6 ) . ] . additional phase errors may be incurred because of uncertainties in the epoch , or even in the geometric elements @xmath43 ( eccentricity ) and @xmath12 , which essentially determine the location of the secondary minimum in relation to the primary through the quantity @xmath44 . 1 shows some hint of this effect ( lower panel ) . as pointed out by m00 , it is to be expected that some measurements will fall on the branches of the minima , and therefore they will underestimate the true amplitude . to assess the importance of this effect they plotted the measured eclipse depths as a function of phase ( their fig . 6 ) . from the fact that not all low amplitude values showed large displacements from the centers of the eclipses they concluded that the effect was unimportant . if the phases are computed instead with the new ephemeris , the general impression from such a plot is rather different ( see fig . 5 ) : nearly all low amplitude values are quite far from mid - eclipse , and several are even beyond the true limits of the eclipse ( c.f . fig . 1 ) . this suggests that some of the amplitudes may indeed be underestimated , and again that the contamination from measurements outside of eclipse can not be ignored . this prompted us to revise the compilation of eclipse depths by m00 by re - examining all the original data using the new ephemeris . as a result , our list of amplitudes for the primary and for the secondary eclipse is somewhat different : some estimates were removed because they lie outside of eclipse , and new ones were added . the 591 harvard measurements used by m00 were made available to us by the senior author , and in addition we considered the measurements from wachmann ( 1936 ) , nekrasova ( 1938 ) , kordylewski et al.(1961 ) , and mossakovskaya ( 1993 ) . for each source we first computed the average magnitude outside of eclipse . each measurement during an eclipse was then compared to this average level to derive the amplitude . in the case of the harvard data three observations during primary minimum and one during secondary minimum were much _ brighter _ than the average level outside of eclipse ( by up to 0.75 mag ) , and were rejected . on nights with several brightness estimates during eclipse we averaged the data as well as the dates , except in cases when the sequence of measurements crossed over from the descending branch to the ascending branch . in such instances the average brightness will always underestimate the true amplitude at the average phase , and therefore we split the data at the center of the eclipse . the uncertainty we assigned to an amplitude estimate is the standard deviation of a single observation outside of eclipse , for each source . in cases where several measurements were averaged , the error was decreased by the square root of the number of observations . finally , we included an estimate in 1991 derived from the hipparcos measurements ( esa 1997 ) , which was shown by ts00 to be very close to mid - secondary eclipse . table 1 and table 2 list all the measurements we considered for the primary ( 40 ) and the secondary eclipse ( 45 ) , respectively . [ tab1 ] [ tab2 ] unfortunately the original dates of the measurements by lehmann ( 1991 ) have not been published , and this prevents us from recomputing the phases to establish whether they can be incorporated into the analysis . we have therefore chosen not to use them here . the procedure we followed consists of fitting the @xmath45 model ( composed of @xmath20 and @xmath46 for the primary and secondary eclipse , and @xmath11 , as described above ) to the eclipse depth measurements , in order to determine @xmath11 . measured amplitudes for both the primary and secondary eclipses were used simultaneously . because many of these estimates are not exactly centered at the minima , and therefore will always underestimate the true amplitude , they must first be corrected for this systematic effect . this was done as follows . for each date of observation we predicted an inclination angle using a preliminary estimate of @xmath11 . with this we computed a synthetic light curve using ebop , which therefore has an amplitude appropriate for the date of the measurement . next we calculated the phase of the observation using the ephemeris , and we applied a correction to the measured depth equal to the difference in magnitude between a point on the curve with the same phase as the observation , and a point at the center of the eclipse . amplitude measurements displaced in phase by 0.017 or more from the center of a primary or secondary minimum ( half width = 0.02 ) were not considered , to guard against possible errors in the ephemeris ( period or epoch , or perhaps even residual variations that have not been accounted for caused , for example , by the presence of additional stars in the system ) and also because such a small drop in brightness is essentially within the errors of measurement of the visual and photographic techniques . we then carried out the fit for @xmath25 and @xmath23 , and improved our preliminary estimate of @xmath11 . the procedure is iterative , and at each step the corrections are slightly different . the eclipse ephemeris reported by ts00 was derived by those authors simultaneously with the apsidal motion and with the spectroscopic elements of the eclipsing binary and of the tertiary star ( including light - time effects , which can amount to 1/10 of the eclipse half - width ) . because the apsidal motion depends on the assumed inclination angle , and this angle is time - dependent , ts00 adopted their own determination of @xmath11 to account for this variation . therefore , the phases we initially compute here from this ephemeris also depend in principle on that determination of @xmath11 relied to some extent on observations by tashpulatov ( 1965 ) , through the estimated epoch of cessation of eclipses . their ephemeris derivation indirectly incorporated this time dependence of @xmath10 , although in practice it has little or no effect on the results . as mentioned in 3 , the tashpulatov observations are now considered to be unreliable however , because of the procedure adopted here , any residual effect from those data on the present analysis disappears completely after the first iteration . ] . in order to be self - consistent , for the present analysis we have recomputed the ephemeris at each iteration using the same methods and the same data employed by ts00 ( times of eclipse , and radial velocities ) , starting from their estimate . successive iterations use updated expressions for @xmath11 , and the change in the phases calculated at each step was found to be very small . the apsidal motion and spectroscopic elements also did not change significantly . the procedure converged in 7 iterations , using standard non - linear least - squares techniques at each step to solve for @xmath25 and @xmath23 ( e.g. , press et al . 1992 ) . in order to ensure that we converged at the absolute minimum of the @xmath47 surface , rather than at a local minimum , we checked the results using a genetic algorithm ( see charbonneau 1995 ) , which explores all of parameter space . 6 displays our best fit to the amplitude measurements of the primary and secondary eclipse , in which the resulting time variation of the inclination angle is represented by @xmath48 $ ] . the maximum amplitude ( corresponding to central eclipses ) according to this fit was therefore reached somewhere between 1904 and 1907 . table 1 and table 2 include the phase displacement of each measurement , the final amplitude corrections we applied , and the corrected eclipse depths as used in the least squares adjustments and figures . as pointed out earlier , a linear approximation to the time dependence of the inclination angle may be valid over relatively short time intervals ( i.e. , the duration of the observations in our case ) , but in reality the change in @xmath10 is more complex . perturbations induced by the third star on the inner pair cause the angular momentum vectors of both orbits to precess around the total angular momentum vector of the system , which is fixed in space . the phenomenon is known as the regression of the nodes " effect ( sderhjelm 1975 ; mazeh & shaham 1976 ) . one of the observable consequences is a periodic change in the inclination angle of the inner orbit that may be expressed conveniently for our case as @xmath49,\ ] ] in which @xmath50 is the angle between the invariable plane of the system ( which is perpendicular to the total angular momentum vector ) and the plane of the sky . the angle @xmath51 is measured between the orbital plane of the eclipsing binary and the invariable plane , and the argument of the cosine term on the far right represents the secular change in @xmath52 , the longitude of the ascending node of the binary orbit , with a period @xmath53 . the angle @xmath50 is constant , but in general @xmath51 may also vary on long timescales , although this would be problematic to measure and is likely to be a small effect . for the purpose of this paper we will consider @xmath51 not to vary . as before , @xmath23 represents the epoch at which @xmath10 reaches @xmath54 . the angle @xmath55 is the position angle of the ascending node at the same instant . from the condition of central eclipses @xmath55 is given by @xmath56 , and thus the four unknowns in the new expression for @xmath11 are @xmath50 , @xmath51 , @xmath53 , and @xmath23 . given the quality of the data available for ss lac it is unrealistic to attempt to solve for all four parameters . however , the nodal period can be estimated from the theory of three - body interactions . ts00 gave an order - of - magnitude estimate of 500 yr . an improved value may be obtained from the expression derived by sderhjelm ( 1975 ) ( his eq.[27 ] ) , which depends not only on the masses of the three stars ( and therefore on the unknown inclination angles of the inner and outer orbits ) , but also on the orbital periods ( known ) , the outer eccentricity ( known ) , and the relative angle between the orbits as well as the orientation of each orbit relative to the invariable plane of the system . several useful constraints on some of these quantities were discussed by ts00 . for example , from the fact that eclipses no longer occur the inclination angle of the binary at the present time must be smaller than @xmath57 , based on the size and separation of the components . in addition , the inclination of the plane of the outer orbit with respect to the line of sight is expected to be no smaller than @xmath58 based on the spectroscopic solution , since smaller angles would imply that the third star would be massive enough and therefore bright enough that it would have been detected spectroscopically , if it is a main - sequence star . this , in turn , along with the measured apsidal motion , restricts the range of possible values for the relative inclination angle between the orbits ( @xmath59 ) to be between about @xmath60 and @xmath58 ( ts00 ) . though all these limits are helpful , the various angles involved are still not known exactly , so we have performed monte carlo simulations subject to all available constraints in order to determine the probability distribution of @xmath53 . the result is shown in fig . periods near 680 yr seem more likely , but the range of possible values extends from 600 to 700 yr . sderhjelm ( 1982 ) has tested the accuracy of the analytical expression used above for @xmath53 by direct integration of the equations of motion in the three - body problem . for stars with orbital and dynamical characteristics similar to those in ss lac he concluded that @xmath61 may be overestimated by that equation by up to 10% or so . we may therefore adopt as a representative value for ss lac a nodal period of @xmath62 yr . to arrive at the best fit to the observations we again used a genetic algorithm to search the parameter space , and found two solutions that are conjugates of each other ( the result of a reflection of some of the angles around @xmath63 ) . they represent physically equivalent scenarios that correspond to direct and retrograde motion of the binary on the plane of the sky . to determine which of these configurations holds for this system requires astrometric measurements , which are not available for ss lac . in both cases the plane of the orbit of the binary is moving away from @xmath54 at the present time , and the consequence for eclipses is the same . mathematically the ambiguity is expressed by the inclination angle numerically increasing or decreasing from @xmath54 during the 20@xmath0 century . choosing the scenario in which @xmath10 is decreasing ( as was done implicitly in 4 and following the usual convention ) , we performed least - squares fits as in 4 to solve for @xmath50 , @xmath51 , and @xmath23 , using the data for both eclipses simultaneously . the corrections to the measured amplitudes were adopted from the final iteration in 4 ( see table 1 and table 2 ) . we obtained @xmath64 , @xmath65 , and @xmath66 , which are consistent with all available constraints . the angles @xmath50 and @xmath51 are significantly correlated , and this is reflected in their formal uncertainties . 8 shows this fit ( solid lines ) along with the corrected observations . for comparison we show also the solution from 4 using a linear model for @xmath11 ( dashed lines ) , which is seen to be very similar to the new result and provides an equally good fit . the epoch of maximum eclipse amplitude is essentially the same in both solutions . even though the fits from both models for @xmath11 are virtually indistinguishable given the errors in the observations , the corresponding predictions for the long - term behavior of the inclination angle and the implications for the eclipsing seasons of ss lac are very different . according to the linear model the inclination angle completes a full revolution in 2500 yr , leading to eclipse seasons that repeat every 1250 yr . however , a secular change in @xmath10 is not what is generally expected from three - body interactions . instead , the inclination typically oscillates in a roughly sinusoidal fashion , the exact shape depending on the angles involved in eq.(2 ) . the period of this modulation for ss lac is about @xmath67 yr , as discussed above , which is the length of a full nodal cycle . the eclipse seasons " ( defined here as intervals when eclipses happen ) can in general occur once or twice during the cycle . in the present case the indications are that there are _ two _ seasons per nodal cycle , each lasting roughly 100 yr . this is shown in fig . 9 , where we display the depth of the eclipses ( primary and secondary ) over an interval of 1500 yr according to the model fit in 5 , as well as the behavior of the inclination angle to the line of sight . eclipses can take place only for values of @xmath10 in the shaded area indicated in the lower panel ( @xmath68 ; see ts00 ) . the eclipse seasons recur in pairs , which combined last for about 350 yr . the observations during the 20@xmath69 century ( top two panels ) are seen to cover the second season of one of such pairs , and no further eclipses are expected until shortly after the year 2200 , if the present model is correct . for comparison we show also in the bottom panel the trend expected for the inclination angle according to the linear model ( dashed line ; 4 ) , which is seen to match the slope of the more realistic model very well at the present time . the conjugate solution mentioned in 5 gives an identical pattern of eclipse seasons , but the function representing @xmath11 is a reflection about an axis at @xmath18 of the curve drawn in the lower panel of fig the end of an eclipse season depends in practice on the precision of the measurements , as argued by m00 . they concluded on the basis of their analysis of the amplitudes that , in the absence of photoelectric data , the eclipses effectively ended between 1937 and 1938 . the slower decline resulting from the present fits favors a somewhat later date around 1945 for the primary eclipses and @xmath11950 for the secondary eclipses , if we assume that amplitudes under 0.1 mag are undetectable by the visual and photographic techniques applied to ss lac . also , central eclipses occurred some 5 or 6 years earlier than in the m00 estimate according to the fits in 4 or 5 . though probably closer to the truth than the linear form of @xmath11 , the more sophisticated regression of the nodes " model is not without its weaknesses . the fit to this model is not as robust as the simpler approximation , partly because of the lack of amplitude measurements to the left of the maximum . instead of two unknowns , three quantities must be solved for ( after fixing @xmath53 ) out of the limited amount of information contained in the data . in addition , the errors given for @xmath50 , @xmath51 , and @xmath23 do not include uncertainties in the mean light curve adopted from ts00 , in the value of @xmath53 , or in the constraints used on @xmath51 , which are based on the measured apsidal motion as reported by ts00 . all these uncertainties are rather difficult to quantify . we can not rule out that , given additional measurements of the depth of the minima , the detailed shape of the function depicted in fig . 9 as well as the pattern of the eclipse seasons described above could change somewhat , particularly regarding the spacing of the two seasons in each nodal cycle . as it turns out , the inclination angle of the orbital plane of the binary extrapolated to the mean epoch of the spectroscopic observations by ts00 ( @xmath11998 ) remains ill - determined . from the linear model for @xmath11 we obtain @xmath70 , while the other model predicts @xmath71 , which would seem to indicate good agreement . although the formal errors in these quantities are roughly @xmath72 , the true uncertainty is likely to be larger so that the absolute dimensions of the components can still not be computed reliably , especially considering also the possible systematic errors in the mean light curve ( see below ) . the main limitation in the present analysis lies in the quantity and quality of the data , both for the amplitude measurements and the light curves . however , improvements can still be made in both areas . it is possible that useful information on the depths of the eclipses can still be extracted from the harvard plates that were not re - measured by schiller et al . ( 1996 ) , as well as from the valuable plate material obtained at the sonneberg observatory and used by lehmann ( 1991 ) , but never published . the light curve solutions might also benefit from these new measurements , and this is important not only to improve our knowledge of the physical properties of the components in ss lac , but for the determination of the depth functions @xmath73 and @xmath74 as well ( 2.1 ) . the most difficult problem faced by all previous investigators attempting to fit the light curves of ss lac has been the fact that the inclination angle of the binary changes with time . inevitably this will affect the solution in subtle ways , and it is likely that several of the photometric elements are biased to some degree as a result of the averaging of the measurements during the minima , as the amplitude changes . a complication of a similar nature has been noted before by other investigators concerning the apsidal motion in eccentric binaries , particularly in systems with short apsidal periods . in this case , even if an accurate apparent period from a linear ephemeris is used for the primary , the effect is a distortion and widening of the secondary minimum because of the changing longitude of periastron . an example where this distortion is quite important is v477 cyg . the solution to the problem of variable photometric elements would seem rather obvious , yet the common practice has been to circumvent the issue by dividing up the data into shorter time intervals , with the effect that data sets are poorer and the phase coverage more incomplete , as in ss lac . if the change in the element(s ) is linear to first order , as is often the case , it should not be difficult to correct for this directly in the computer program used to solve the light curve , allowing for the variation as a function of time . in the apsidal motion case , this has in fact been done for the system mentioned above , v477 cyg , by gimnez & quintana ( 1992 ) using the computer program ebop . they modified the appropriate subroutines so that the phase calculated for each observation accounts for the change in the longitude of periastron using a pre - determined value of @xmath75 . similarly , either ebop , the wilson - devinney code , or other programs in use could in principle be modified to account for @xmath25 in calculating the theoretical flux to be compared with each photometric measurement , thus referring all observations to a standard ( arbitrary ) epoch with a _ fixed _ inclination angle . more ambitious investigators might be tempted to add @xmath25 as a variable to be determined from the solution along with the other light elements , but this may only be possible in the best observed systems . alternatively , the individual photographic and visual observations for ss lac may first be corrected along the lines described in 4 for the amplitude measurements , and then submitted to the light curve analysis . this would require no change in the current computer codes . the present analysis of the eclipse amplitude measurements could then be repeated once an improved light curve solution becomes available , and the procedure iterated until convergence . among the very few eclipsing binaries in which the amplitudes of the minima have been seen to change , to the point where the eclipses disappear completely due to the cumulative effect of the perturbations from a third star in the system , only a handful are as well documented as ss lac . despite obvious limitations in the observational material spanning more than a century , this has presented us with a rare opportunity to learn more about the configuration of the system in the light of current theoretical ideas about three - body interactions among stars . in particular , the eclipse amplitude measurements provide a valuable record of the change in the inclination angle that , with the possible exception of the succinct report by lehmann ( 1991 ) , has not been investigated in detail by applying the direct constraints offered by these observations . the three main contributions of this paper to the study of this phenomenon are ( i ) the use of a more accurate ephemeris ( from ts00 ) to select from the historical record the brightness measurements that were obtained during an eclipse ; ( ii ) the application of corrections to the amplitude estimates to account for the systematically lower values typically measured on the branches of the minima ; and ( iii ) the consideration of more physically realistic models than used before for the change in the eclipse depth , based on the known properties of the system ( mean light curve solution ) and expressions for the expected time variation of the inclination angle . for the latter we have explored both a linear approximation that is found to be quite satisfactory over the period covered by the observations , and a more rigorous prescription derived from the theory of the regression of the nodes " effect in triple systems . equally good fits are achieved with both expressions for @xmath11 , but realistic long - term predictions on the future occurrence of eclipses in ss lac can only be made from the second approach . our fits suggest that the eclipse seasons ( duration @xmath1100 yr ) come in pairs that repeat after about 600 yr , and that we may expect the current lack of eclipses to last until approximately the beginning of the 23@xmath76 century . we are grateful to the referee , dr . e. f. milone , for a number of helpful comments as well as for suggesting the use of , and generously providing the individual measurements of the harvard patrol camera plates made by schiller et al . this research has made use of the simbad database , operated at cds , strasbourg , france , and of nasa s astrophysics data system abstract service . schiller , s. j. , bridges , d. , & clifton , t. 1996 , in the origins , evolutions , and destinies of binary stars in clusters , asp conf . ser . , 90 , eds . e. f. milone & j .- c . mermilliod , ( san francisco : asp ) , 141	eclipses in the 14.4 day period double - lined binary ss lac were observed photographically and visually early in the 20@xmath0 century , but stopped some 50 or 60 years ago . this has been explained by the presence of a distant third star in the system , which has now been detected spectroscopically with a period of 679 days . the plane of the orbit of the binary is changing relative to the line of sight in response to perturbations from this third object . a recent analysis by milone et al . of all photometric material available for the system , including a re - measurement of original harvard plates , has confirmed earlier reports of changes in the depth of the eclipses as a function of time , which are due to the third star . in this paper we discuss our detailed analysis of the eclipse amplitude measurements , and extract from them information on the change in the inclination angle of the binary over the last century . our use of a much improved ephemeris for the system by torres & stefanik was found to be crucial , and prompted us to re - determine all the amplitudes from the historical data at our disposal , including the harvard material used by m00 . systematically lower measurements on the branches of the minima were properly accounted for , and we made use of both a linear approximation to the time variation of the inclination angle and a more realistic model based on the theory of three - body interactions ( regression of the nodes " effect ) . the nodal cycle is found to be @xmath1600 yr , within which _ two _ eclipse seasons " occur , each lasting about 100 yr . the non - eclipsing status of the system is expected to continue until the beginning of the 23@xmath2 century . to appear in _ the astronomical journal _ , april 2001
in this paper we derive explicit asymptotically sharp estimates for the constant of hlder continuity in the spherical metric of quasiconformal ( qc ) mappings . these results are based on reducing the problem to the euclidean metric case , where hlder continuity is well - known . the paper @xcite provides explicit estimates with some nice properties ( cflemma [ m1fv ] ) that will be shown to hold for the constants in the spherical metric also . a similar , but more restricted , result has previously appeared as lemma 4.1 of @xcite ( cited as lemma [ bonlem ] in this paper ) . this result gives an estimate for the hlder constant with respect to the spherical metric of planar quasiconformal mappings , which , however , is not asymptotically sharp as @xmath5 . this paper is organized as follows : in the next section the main results are stated , in section 3 the standard notation is reviewed and some explicit estimates pertaining to qc mappings are presented . section 4 contains various lemmas and the proof of the main results . in section 5 , an explicit bound for the hlder constant in the spherical metric of @xmath2-quasisymmetric real functions is derived . we first introduce some notation and terminology for the formulation of our main results . for a more complete account , see the following section . [ holcdef ] we define @xmath6 ( for @xmath7 ) to be the smallest constant such that the following inequalities hold for all @xmath8 and for all @xmath2-quasiconformal mappings @xmath9 , with @xmath10 , @xmath11 that satisfy the additional conditions indicated : * @xmath12 * @xmath13 * @xmath14 * @xmath15 where @xmath16 and @xmath1 is the spherical metric ( defined in([qmetric ] ) ) . for brevity , we will write @xmath17 , @xmath18 , @xmath19 and @xmath20 . the two last definitions are to be understood as replacing the condition @xmath8 by the condition @xmath21 . we can also define analogous constants for @xmath22 , see section 5 . note that for @xmath23 we require that @xmath24 , whereas for @xmath25 we only have the normalization @xmath26 . the reason for this discrepancy will become clear in the proof of theorem [ m_3(2 ) ] . [ m1fv ] ( @xcite , pp115 - 7 ) the constant @xmath25 has the following properties : * @xmath27 as @xmath28 , * @xmath29 is bounded for fixed @xmath30 , * @xmath29 is bounded for fixed @xmath31 moreover , @xcite provides explicit estimates that exhibit the described behavior and shows that @xmath32 has properties ( 1 ) and ( 3 ) , and provides an explicit estimate . in terms of these constants we may state bonfert - taylor s result as [ bonlem ] ( lemma 4.1 , @xcite ) @xmath33 . note that this constant does not satisfy the most important of the properties above , property ( 1 ) . however , this result proves property ( 2 ) for @xmath34 , but not property ( 1 ) , which is perhaps the most important . the main results of this paper show that ( 1 ) , ( 2 ) and ( 3 ) hold for @xmath23 and ( 1 ) and ( 2 ) hold for @xmath35 . @xcite also shows that @xmath32 does not satisfy ( 3 ) . unfortunately , we were unable to settle whether @xmath35 satisfies ( 2 ) . our results are stated in the following two theorems : [ annthm ] let @xmath36 denote the grtzsch constant ( see @xcite , ( 8.38 ) ) , @xmath37 be as in lemma [ etaest ] and @xmath38 be as in lemma [ annlem ] . then @xmath39 where @xmath40 moreover , @xmath38 , and therefore @xmath35 , is asymptotically sharp for @xmath5 with this choice of @xmath41 . in particular , for @xmath42 , @xmath35 satisfies the quite crude estimate : @xmath43 this means that @xmath44 for @xmath42 . if we assume instead that @xmath45 we have the better estimates @xmath46 [ m_3(2 ) ] the constant @xmath23 satisfies ( 2 ) in lemma [ m1fv ] . in particular , @xmath47 we denote by @xmath48 the standard basis of @xmath49 . the following standard notation will be used : @xmath50 @xmath51 we define the spherical metric @xmath1 by means of the canonical projection onto the riemann sphere , hence , for @xmath52 , @xmath53 we denote the dimension of the euclidean space under consideration by @xmath30 . till section 5 , it is assumed that @xmath54 . the letter @xmath2 will denote the constant of quasiconformality of the arbitrary quasiconformal mapping @xmath55 ( of @xmath56 ) . more precisely , this entails the following assumptions : @xmath57 and @xmath58 is an acl homeomorphism , satisfying @xmath59 a.e . in @xmath56 , where @xmath60 stands for the jacobian ( cf@xcite , theorem 9.9 ) . for ease of reference we define the classes @xmath61 and refer to mappings of this class as normalized @xmath2-qc mappings . note that for @xmath62 , @xmath63 consists of isometries only , that is , reflections in planes containing the @xmath64-axis and rotations about the same , by the constancy of the cross ratio under mbius mappings . from the definition it also follows that if @xmath65 then @xmath66 . let @xmath67 be a homeomorphism . let @xmath68 and @xmath69 be domains in @xmath49 . we say that a mapping @xmath70 is @xmath71-quasisymmetric if @xmath72 for all @xmath73 with @xmath74 . [ etaest ] ( @xcite theorems 14.6 and 14.8 , @xcite theorem 7.47 ) let @xmath36 as in theorem [ annthm ] . for @xmath54 and @xmath3 there exists a bijection @xmath75 such that every @xmath2-qc mapping of @xmath49 is @xmath37-qs , and @xmath37 satisfies the following bounds : * @xmath76 * @xmath77 , for @xmath78 , * @xmath79 , for @xmath80 , * @xmath81 , for @xmath82 , * @xmath83 , for @xmath84 for the remainder of this paper let us write @xmath85 and @xmath86 . it is well - known that @xmath87 and that @xmath88 $ ] ( see @xcite , p. 169 ) . [ etaestimp ] ( @xcite , 14.36 ( 4 ) ) for @xmath89 , we have the better estimate @xmath90 the main results , theorems [ annthm ] and [ m_3(2 ) ] , will be proved in this section . to accomplish this , we first temporarily turn to a more general setting . suppose , therefore , that @xmath91 is an arbitrary @xmath92-bihlder continuous mapping with constant @xmath93 ( with respect to the euclidean metric ) . what can we say about the hlder continuity of @xmath91 with respect to the spherical metric ? in this general setting we may state the following lemma , which turns out to be asymptotically sharp for qc - mappings : [ holderlem ] let for @xmath94 consider an arbitrary @xmath95bihlder continuous mapping , @xmath70 , fixing the origin ( which entails that @xmath68 and @xmath69 contain it ) . that is to say , there exists a @xmath96 such that @xmath97 for all @xmath98 . then @xmath91 is hlder continuous with respect to the spherical metric as well , in particular : @xmath99 where @xmath100 . by ( [ qmetric ] ) , the second inequality of ( [ bihe ] ) is equivalent to @xmath101 and hence @xmath102 , where @xmath103 and the supremum is taken over @xmath104 and over all mappings @xmath91 that satisfy the assumptions . using the first inequality of ( [ bihe ] ) we conclude , by setting @xmath105 in ( [ bihe ] ) , that @xmath106 then , since @xmath96 and @xmath107 , we have @xmath108 to prove that @xmath109 we take the logarithmic partial derivative with respect to @xmath110 of both sides . since the left and right hand side of ( [ 1.5c ] ) are equal when @xmath111 , we only need to prove that the derivative of the left hand side is greater than that of the right hand side for @xmath112 that is , we need to show that @xmath113 where @xmath114 . the left hand side has an infimum greater than @xmath115 and the right hand side the supremum @xmath116 , so we are done . note that a mapping can be @xmath110-bihlder continuous in the sense of ( [ bihe ] ) for @xmath117 only if @xmath68 and @xmath69 are bounded . [ qhcbn ] @xmath118 follows directly from lemma [ holderlem ] as @xmath55 is @xmath119-hlder continuous in @xmath120 with constant @xmath25 by definition . from the lemmas [ m1fv ] and [ holderlem ] we derive the properties ( 1)-(3 ) for @xmath121 . it does not provide the hlder continuity throughout space . it is possible to remedy this shortcoming as will be shown , however , in doing so , we will loose property ( 2 ) . therefore we first give an independent proof of property ( 2 ) . by lemma [ holderlem ] this is clear if @xmath122 or @xmath123 . let @xmath124 and @xmath125 and @xmath126 be given . we will prove that there exists a point @xmath127 in the intersection of the line through @xmath128 and @xmath129 with @xmath130 such that @xmath131 . let @xmath132 be the image of @xmath128 in the canonical projection of @xmath56 onto @xmath133 etc . then , by definition , @xmath134 and so on . let @xmath135 be the radius of the image of the line containing @xmath128 and @xmath129 under the projection . of the two possible @xmath127 s , we chose the one on the shorter arc joining @xmath132 and @xmath136 . then we need to show that @xmath137 let us differentiate the left - hand - side with respect to @xmath119 . we the find easily that the left - hand - side has a maximum at @xmath138 . we then have to show that @xmath139 . now this follows directly from the equation @xmath140 . from lemma [ holderlem ] it follows that @xmath141 here we have used the fact that a continuous mapping has the same hlder constant for @xmath120 and @xmath142 . we combine the previous estimates and conclude that @xmath143 @xmath144 @xmath145 thus @xmath146 . note that this estimate is not asymptotically sharp . we now return to deriving an asymptotically sharp estimate and start by presenting an auxiliary lemma stating an explicit relationship between the constant and the domain of definition of the mapping : [ annlem ] for @xmath7 @xmath147 let @xmath126 and @xmath8 . define an auxiliary mapping @xmath91 by : @xmath148 since @xmath149 it follows from the definition of @xmath25 that @xmath150 which is equivalent to @xmath151 and thus , by lemma [ etaest ] , ( [ expannest ] ) holds with the constant @xmath152 indicated . now we will derive a new version of lemma [ holderlem ] , which incorporates the variability of @xmath152 in the proof and is thus specialized to quasiconformal mappings . [ holderlemqc ] let @xmath126 . then @xmath153 is hlder continuous with respect to the spherical metric with constant @xmath154 satisfying @xmath155 this is proved as lemma [ holderlem ] except that we will use different methods of estimating the upper bounc @xmath156 of @xmath157 . if @xmath158 we use the estimate @xmath159 in this case we easily see that @xmath160 . for @xmath161 we have , by the hlder continuity of @xmath162 , @xmath163 . then we proceed exactly as lemma [ holderlem ] to get @xmath164 . it follows that @xmath165 recall that in lemma [ holderlem ] the hlder constant in the spherical metric was shown to be less than @xmath166 so we are done . we are now in a position to derive easily the main result . by lemmas [ annlem ] and [ holderlemqc ] the estimate holds for @xmath167 . since the inversion @xmath168 is a q - isometry , we conclude that the estimate holds also for @xmath169 . we now deal with the remaining case : @xmath170 . then @xmath171 the last inequality holds , since we may assume ( otherwise inequality ( [ expm_4est ] ) is trivial ) @xmath172 as @xmath173 . from this it follows immediately that @xmath174 which concludes the proof of ( [ expm_4est ] ) . for asymptotical sharpness we still need to prove that @xmath175 when @xmath5 for @xmath41 as given in ( [ rdef ] ) , i.e. that @xmath176 at the limit . this follows from the following chain of inequalities : @xmath177 thus we see that the bound of @xmath35 is asymptotically sharp . note how using this latter method does not give us a better explicit bound for @xmath23 than for @xmath35 . this is a consequence of our operating in euclidean space , where the sharper restriction of the mappings on the unit circle is not much different from only fixing unity as viewed from near infinity . also note that already for quite small values of @xmath2 , the estimate in theorem [ annthm ] is worse that that in the proof of theorem [ m_3(2 ) ] . if @xmath2 is close to one , we may derive the following bound for @xmath178 in terms of our new constant @xmath179 : @xmath180 it is valid if @xmath181 . in particular , @xmath182 if @xmath183 . these claims are proved exactly as lemma [ holderlem ] so the proofs are omitted here . we see that from bounds on the hlder constants in the spherical metric we can derive bounds for the constants in the euclidean metric , as well as the other way around . in @xcite , a one parameter family of @xmath2-quasisymmetric ( qs ) functions is defined as those functions @xmath184 that satisfy @xmath185 for all @xmath128 and @xmath186 . we will now derive an explicit bound for the hlder constant in the spherical metric of this class of functions . the proof of this is similar to the existence proof given by lehto and virtanen ( @xcite , p. 57 ) , except that the following lemma will be used : [ lehlem](@xcite ) let @xmath55 be @xmath2-qs in the sense of @xcite with constant @xmath2 . then @xmath55 can be extended to an @xmath187-qc mapping @xmath188 where @xmath189 and @xmath190 denotes the upper half - plane . the point @xmath128 lies in a region bounded by a line and a circle , as shown in figure 1a . this means that @xmath199 lies in a region bounded by two spheres , as shown in figure 1b . it is immediately clear that @xmath129 is as far from the origin as any other point in the region , so that @xmath200 . we calculate @xmath201 from @xmath202 and @xmath203 ( note that here @xmath119 refers to an angle as indicated in the picture , not to @xmath204 . ) we subtract the second equation from the first to get @xmath205 now using this in the first equation gives us the desired formula @xmath206 . denote the symmetric @xmath212-qc extension of @xmath55 to a mappin from @xmath191 to @xmath191 by @xmath55 , again ( here we use lemma [ lehlem ] ) . we assume without loss of generality that @xmath11 and @xmath213 . since @xmath55 is @xmath212-qc , @xmath214 . we may assume without loss of generality that @xmath215 . let @xmath216 be a point , such that the projection of @xmath217 in the real axis equals @xmath218 . then @xmath219 where the last inequality holds , since @xmath220 is increasing and @xmath221 , since @xmath222 . therefore we may take @xmath223 in lemma [ auxlem ] to conclude that @xmath224 with @xmath196 the inversion in @xmath197 . let @xmath225 . since @xmath196 is a 1-qc , @xmath91 is @xmath212-qc . since @xmath55 fixes @xmath226 , @xmath91 fixes @xmath227 . however , ( [ zero ] ) means that @xmath228 , since @xmath229 . let @xmath230 be a mbius transformation with @xmath231 and @xmath232 . then , since @xmath91 is symmetric in @xmath227 , @xmath233 . now @xmath230 is bilipschitz with constant @xmath234 ( see eg@xcite , chapter 1 , for these elementary mbius mapping results . ) since @xmath235 is @xmath212-qc and fixes @xmath227 , 0 and @xmath236 , we have @xmath237 , for @xmath238 by theorem 15.5 of @xcite . it follows from the bilipschitz property of @xmath230 that @xmath239 when we set @xmath240 and @xmath241 , we get @xmath242 for @xmath243 , which is what we wanted to show . [ conj ] the previous theorem could be extended to higher dimensions by using the extension result of @xcite instead of that of @xcite , which would allow us to derive estimates similar to those in the begining of this paper . this does not , however , give the explicit results that we have strived for .	this paper contains bounds for the distortion in the spherical metric , that is to say bounds for the constant of hlder continuity of mappings @xmath0 where @xmath1 denotes the spherical metric . the mappings considered are @xmath2-quasiconformal ( @xmath3 ) and satisfy some normalizations or restrictions . all bounds are explicit and asymptotically sharp as @xmath4 .
-5 mm let @xmath2 be a standard brownian motion taking values in @xmath3 and let @xmath4 $ ] denote the random set @xmath5 = \{b_r : s \leq r \leq t\ } .$ ] for @xmath6 , we say that @xmath2 is a * _ cut point _ for @xmath7 $ ] if @xmath8 = \emptyset $ ] ; * _ frontier point _ for @xmath7 $ ] if @xmath2 is on the boundary of the unbounded component of @xmath9 $ ] ; * _ pioneer point _ for @xmath7 $ ] if @xmath2 is on the boundary of the unbounded component of @xmath10 $ ] , i.e. , if @xmath2 is a frontier point for @xmath11 $ ] . i will discuss the following result proved by oded schramm , wendelin werner , and myself . * theorem 1 . * @xcite _ if @xmath2 is a standard brownian motion in @xmath3 , then with probability one , @xmath12\ ; ] = 3/4 , \ ] ] @xmath13\ ; ] = 4/3 , \ ] ] @xmath14\;]= 7/4 , \ ] ] where @xmath15 denotes hausdorff dimension . _ mandelbrot @xcite first gave the conjecture for the brownian frontier , basing his conjecture on numerical simulation and then noting that simulations of the frontier resembled simulations of self - avoiding walks . it is conjectured that the scaling limit of planar self - avoiding walks has paths of dimension @xmath0 . duplantier and kwon @xcite used nonrigorous conformal field theory techniques to make the above conjectures for the cut points and pioneer points . more precisely , they made conjectures about certain exponents called the brownian or simple random walk _ intersection exponents_. more recently , duplaniter @xcite has given other nonrigorous arguments for the conjectures using quantum gravity . to prove theorem 1 , it suffices to find the values of the brownian intersection exponents . in fact , before theorem 1 had been proved , it had been established @xcite that the hausdorff dimensions of the set of cut points , frontier points , and pioneer points were @xmath16 and @xmath17 , respectively , where @xmath18 are defined by saying that as @xmath19 , @xmath20 \cap b[\frac12 + \epsilon^2 , 1 ] = \emptyset \ } \approx \epsilon^{\eta_1},\ ] ] @xmath20 \cup b[\frac{1}{2 } + \epsilon^2,1 ] \mbox { does not disconnect } b_{1/2 } \mbox { from infinity } \ } \approx \epsilon^{\eta_2},\ ] ] @xmath21 \mbox { does not disconnect } 0 \mbox { from infinity } \ } \approx \epsilon^{\eta_3}.\ ] ] it had also been established @xcite that the analogous exponents for simple random walk are the same as for brownian motion . there are two main ideas in the proof . the first is a one parameter family of conformally invariant processes developed by oded schramm @xcite which he named the stochastic loewner evolution ( @xmath1 ) . the second is the idea of `` universality '' which states roughly that all conformally invariant measures that satisfy a certain `` locality '' or `` restriction '' property must have the same exponents as brownian motion ( see @xcite ) . in this paper , i will define @xmath1 and give some of its properties ; describe how analysis of @xmath1 leads to finding the brownian intersection exponents ; and finally describe some other planar lattice models in statistical physics at criticality that can be understood using @xmath1 . -5 mm i will give a brief introduction to the stochastic loewner evolution ( @xmath1 ) ; for more details , see @xcite . let @xmath22 denote a standard one dimensional brownian motion . if @xmath23 and @xmath24 is in the upper half plane @xmath25 , let @xmath26 be the solution to the loewner differential equation @xmath27 for each @xmath28 , the solution @xmath26 is defined up to a time @xmath29 $ ] . let @xmath30 . then @xmath31 is the unique conformal transformation of @xmath32 onto @xmath33 with @xmath34 as @xmath35 . in fact , @xmath36 it is easy to show that the maps @xmath31 are well defined . it has been shown @xcite that there is a ( random ) continuous path @xmath37 such that @xmath32 is the unbounded component of @xmath38 $ ] and @xmath39 . the conformal maps @xmath31 or the corresponding paths @xmath40 are called the _ chordal stochastic loewner evolution with parameter @xmath41 ( chordal @xmath42)_. it is easy to check that the distribution of @xmath42 is invariant ( modulo time change ) under dilations @xmath43 . using this , we can use conformal transformations to define chordal @xmath42 connecting two distinct boundary points of any simply connected domain . this gives a family of probability measures on curves ( modulo reparametrization ) on such domains that is invariant under conformal transformation . chordal @xmath42 can also be considered as the only probability distributions on continuous curves ( modulo reparametrization ) @xmath44 with the following properties . * @xmath45 as @xmath46 , and @xmath47 for all @xmath48 where @xmath32 is the unbounded component of @xmath49.$ ] * let @xmath50 be the unique conformal transformation with @xmath51 . then the conditional distribution of @xmath52 , given @xmath53 $ ] is the same as the original distribution . * the measure is invariant under @xmath54 there is a similar process called _ radial @xmath42 _ on the unit disk . let @xmath22 be as above , and for @xmath24 in the unit disk @xmath55 , consider the equation @xmath56 let @xmath57 be the set of @xmath58 for which @xmath26 is defined . it can be shown that there is a random path @xmath59 , such that @xmath57 is the component of @xmath60 $ ] containing the origin ; @xmath61 ; and @xmath31 is a conformal transformation of @xmath57 onto @xmath55 with @xmath62 . we can define radial @xmath42 connecting any boundary point to any interior point of a simply connected domain by conformal transformation . the qualitative behavior of the paths @xmath63 varies considerably as @xmath41 varies , although chordal and radial @xmath42 for the same @xmath41 are qualitatively similar . the hausdorff dimension of @xmath53 $ ] for chordal or radial @xmath64 is conjectured to be @xmath65 . this has been proved for @xmath66 see @xcite , and for other @xmath41 it is a rigorous upper bound @xcite . for @xmath67 , the paths @xmath63 are simple ( no self - intersections ) and @xmath68 is a subset of @xmath33 or @xmath55 . for @xmath69 , the paths have double points and hit @xmath70 or @xmath71 infinitely often . if @xmath72 , the paths are space filling . investigation of @xmath42 requires studying the behavior of @xmath42 under conformal maps . suppose @xmath73 is a compact subset of @xmath74 not containing the origin such that @xmath75 and @xmath76 is simply connected . let @xmath77 denote the conformal transformation of @xmath78 onto @xmath33 with @xmath79 . let @xmath63 denote a chordal @xmath42 starting at the origin , and let @xmath80 be the first time @xmath81 that @xmath82 . for @xmath83 , let @xmath84 . let @xmath85 be the conformal transformation of the unbounded component of @xmath86 $ ] onto @xmath33 with @xmath87 as @xmath35 ; define @xmath88 by @xmath89 then @xmath90 satisfies the modified loewner equation @xmath91 for some @xmath92 . in fact @xmath93 . using the loewner differential equation and it s formula , we can write @xmath92 as a local semimartingale , @xmath94 ; here @xmath95 and @xmath88 are random depending on @xmath96 . for @xmath97 , and only @xmath98 , the drift term @xmath95 disappears and hence @xmath92 is a time change of brownian motion . * locality property for @xmath99 . * @xcite _ if @xmath97 , @xmath100 , has the same distribution as a time change of @xmath99 . _ for other values of @xmath41 , the image @xmath101 , has a distribution that is absolutely continuous with respect to that of ( a time change of ) @xmath42 . this follows from girsanov s theorem ( see , e .. g , ( * ? ? ? * theorem i.6.4 ) ) that states roughly that brownian motions with the same variance but different drifts give rise to absolutely continuous measures on paths . similarly , radial @xmath42 can be obtained from chordal @xmath42 by considering its image under a map taking @xmath33 to @xmath55 . for all values of @xmath41 we get absolutely continuous measures ( which is why radial @xmath42 is qualitatively the same as chordal @xmath42 ) , but for @xmath97 we get a special relationship ( * ? ? ? * theorem 4.1 ) . one of the reasons that @xmath42 is useful is that `` crossing probabilities '' and `` critical exponents '' for the process can be calculated . the basic idea is to relate an event about the planar path @xmath63 to an event about the driving process @xmath102 and then to use standard methods of stochastic calculus to relate this to solutions of partial differential equations . as an example , consider chordal @xmath99 in the upper half plane @xmath33 going from @xmath103 to infinity . let @xmath80 be the first time @xmath81 that @xmath104 \cap [ 1,\infty)$ ] ; since @xmath105 , @xmath106 with probability one . let @xmath107 be the event that @xmath108 $ ] , and let @xmath109 be the unbounded component of @xmath110 $ ] . let @xmath111 be the minimum and maximum of @xmath112 \cap \r$ ] ; on the event @xmath113 , @xmath114 and @xmath115 . let @xmath116 denote the _ @xmath117-extremal distance _ between @xmath118 $ ] and @xmath119 $ ] in @xmath120 , i.e. , the number @xmath116 such that @xmath120 can be mapped conformally onto @xmath121 \times [ 0,\pi]$ ] in a way that @xmath118 $ ] and @xmath119 $ ] are mapped onto the vertical boundaries . in order to relate @xmath99 to intersection exponents for brownian motion one needs to understand the behavior of @xmath122 $ ] as @xmath123 for @xmath124 . it is not hard to show that this quantity is closely related to @xmath125 $ ] . if we differentiate ( [ lde.chord ] ) with respect to @xmath24 we get an equation for @xmath126 , and standard techniques of stochastic calculus can be applied to give a differential equation for the function @xmath127 $ ] . we get an exact solution in terms of hypergeometric functions ( * ? ? ? * theorem 3.2 ) . if @xmath128 , so that @xmath129 $ ] , we get the formula given by cardy @xcite for crossing probabilities of percolation clusters ( see [ percsec ] ) . -5 mm as already mentioned , computation of dimensions for many exceptional sets for brownian motion reduces to finding the brownian intersection exponents . these exponents , which can be defined in terms of crossing probabilities for non - intersecting paths , were studied in @xcite . in these papers , relations were given between different exponents and a `` universality '' principle was shown for conformally invariant processes satisfying an additional hypothesis ( the term completely conformally invariant was used there ) . heuristic arguments indicated that self - avoiding walks and percolation should also satisfy this hypothesis . unfortunately , from a rigorous standpoint , we had only reduced a hard problem , computing the brownian intersection exponents , to the even harder problem of showing conformal invariance and computing the exponents for self - avoiding walks or critical percolation . at the same time schramm @xcite was completing his beautiful construction of @xmath42 and conjecturing that @xmath99 gave the boundaries of critical percolation clusters . while he was unable to prove that critical percolation has a conformally invariant limit , he was able to conclude that if the limit was conformally invariant then it must be @xmath99 . the identification @xmath98 was determined from rigorous `` crossing probabilities '' for @xmath42 ; only @xmath97 was consistent with cardy s formula ( see [ percsec ] ) or even the simple fact that a square should have crossing probability @xmath130 . since both brownian motion and @xmath99 were conjectured to be related to the scaling limit of critical percolation , it was natural to try to use @xmath99 to prove results about brownian motion ( and , as mentioned before , the hausdorff dimension of exceptional sets on the path ) ; see @xcite . there were two major parts of the proof . first , the locality property for @xmath99 was formulated and proved ; this allowed ideas as in @xcite to show that the exponents of @xmath99 can be used to find the exponents for brownian motion . second , the exponents for @xmath99 had to be computed . the basic idea is discussed at the end of the last section . what makes @xmath1 so powerful is that it reduces problems about a two - dimensional process to analysis of a one - dimensional stochastic differential equation ( and hence a partial differential equation in one space variable ) . the universality in these papers was in terms of exponents . we now know that the paths of planar brownian motion and @xmath99 are even more closely related . the `` hull '' generated by an @xmath99 is the same as the hull generated by a brownian motion with oblique reflection ( see @xcite ) . in particular , the frontiers ( outer boundaries ) of the two processes have the same dimension . there are now direct proofs that the hausdorff dimension of the frontier of @xmath99 is @xmath0 ( @xcite ) and this stronger universality principle implies the same holds for brownian paths . suppose each vertex of the planar triangular lattice is colored independently white or black , with the probability of a white being @xmath130 . this is called _ critical percolation _ ( on the triangular lattice ) . let @xmath131 be a simply connected domain in @xmath132 and let @xmath133 be disjoint nontrivial connected arcs on @xmath134 . consider the limit as @xmath135 of the probability that in critical percolation on a lattice with mesh size @xmath136 that there is a connected set of white vertices in @xmath131 connecting @xmath137 . it has long been believed that this limit , @xmath138 , exists and is strictly between @xmath139 and @xmath140 . ( note : if the probability of a white vertex is @xmath141 , then @xmath138 is @xmath139 for @xmath142 and @xmath140 for @xmath143 . one of the features of _ critical _ percolation is the fact that this quantity is strictly between @xmath139 and @xmath140 . ) moreover , it has been conjectured that @xmath138 is a conformal invariant @xcite . it is also believed that this limit does not depend on the nature of the lattice ; for example , critical bond percolation in @xmath144 ( each bond is colored white or black independently with probability @xmath130 ) should give the same limit . cardy @xcite used nonrigorous methods from conformal field theory to find an exact formula for @xmath138 ; his calculations were done for @xmath145 and the formula involves hypergeometric functions . carleson noted that the formula was much nicer if one chooses @xmath131 to be an equilateral triangle of side length @xmath140 ; @xmath146 , one of the sides ; and @xmath147 , a line segment of length @xmath148 with one endpoint on the vertex opposite @xmath146 . in this case , cardy s formula is @xmath149 . schramm @xcite went further and , _ assuming existence and conformal invariance of the limit _ , showed that the limiting boundary between black and white clusters can be given in terms of @xmath99 . if @xmath150 denotes the third side of the triangle ( so that @xmath151 is a single point ) , we can consider the limiting cluster formed by taking all the white vertices that are connected by a path of white vertices to @xmath150 . in the limit , the outer boundary of this `` hull '' has the same distribution as the outer boundary of the hull of chordal @xmath99 going from the vertex @xmath152 to the vertex @xmath151 . the identification with @xmath1 comes from the conformal invariance assumption ; schramm determined the value @xmath97 from a particular crossing probability , but we now understand this in terms of the locality property which scaling limits of these boundary curves can be seen to satisfy . cardy s formula ( and generalizations ) were computed for @xmath42 in @xcite . recently smirnov @xcite made a major breakthrough by proving conformal invariance and cardy s formula for the limit of critical percolation in the triangular lattice . as a corollary , the identification of the limit with @xmath99 has become a theorem . this has also led to rigorous proofs of a number of critical exponents for the lattice model @xcite . the basic strategy is to compute the exponent for @xmath99 and to then to use smirnov s result to relate this exponent to lattice percolation . _ loop - erased random walk ( lerw ) _ in a finite set @xmath153 starting at @xmath154 is the measure on self - avoiding paths obtained from starting a simple random walk at the origin , stopping at the first time that it leaves @xmath73 , and erasing loops chronologically from the path . it can also be defined as a nonmarkov chain which at each time @xmath155 chooses a new step using probabilities weighted by the probability that simple random walk starting at the new point avoids the path up to that point ( see , e.g. , @xcite ) . it is also related to uniform spanning trees ; if one choose a spanning tree uniformly among all spanning trees of @xmath73 , considered as a graph with appropriate boundary conditions , then the distribution of the unique self - avoiding path from the origin to the boundary is the same as lerw . wilson gave a beautiful algorithm to generate uniform spanning trees using lerw @xcite . one can hope to define a scaling limit of planar lerw on a domain connecting an interior point to a boundary point by taking lerw on finer and finer grids and taking the limit . there are a number of reasons to believe that this limit is conformally invariant . for example . the limit of simple random walk ( brownian motion ) is conformally invariant and the ordering of points used in the loop - erasing procedure is not changed under conformal maps . also , certain crossing probabilities for lerw can be given by determinants of probabilities for simple random walk ( see @xcite ) , and hence these quantities are conformally invariant . kenyon @xcite used a conformal invariance argument ( using a determinant relation from a related domino tiling model ) to prove that the growth exponent for lerw is @xmath156 ; roughly , this says it takes about @xmath157 steps for a lerw to travel distance @xmath158 . schramm @xcite showed that under the assumption of conformal invariance , the scaling limit of lerw must be radial @xmath159 . he used conformal invariance and a natural markovian - type property of lerw to conclude that it must be an @xmath42 , and then he used kenyon s result to determine @xmath41 . recently , schramm , werner , and i @xcite proved that the scaling limit of loop - erased random walk is @xmath160 . there is another path obtained from the uniform spanning tree that has been called the uniform spanning tree peano curve . this path , which lies on the dual lattice , encodes the entire tree ( not just the path from the origin to the boundary ) . a similar , although somewhat more involved , argument can be used to show that this process converges to the space - filling curve @xmath161 @xcite . a _ self - avoiding walk ( saw ) _ in the lattice @xmath144 is a nearest neighbor walk with no self - intersections . the problem of the saw is to understand the uniform measure on all such walks of a given length ( or sometimes the measure that assigns weight @xmath162 to all walks of length @xmath155 ) . it is still an open problem to prove there is a limiting distribution ; it is believed that such a limit in conformally invariant ( see @xcite for precise statements ) . however , if the conjectures hold there is only one possible limit , @xmath163 . the conformal invariance property leads one to conclude that the limit must be an @xmath42 and @xmath164 is needed in order to have a measure on simple paths . the property that @xmath163 has that is not held by @xmath42 for other @xmath165 is the _ restriction property_. the restriction property is similar to , but not the same , as the locality property . let @xmath73 be a compact subset and @xmath77 the transformation as in [ section 2 ] then @xcite if @xmath166 is an @xmath163 path from @xmath139 to @xmath167 , the distribution of @xmath168 given the event @xmath169 is the same as ( a time change of ) @xmath163 . in fact , the probability that @xmath169 is @xmath170 . if the scaling limit of saw has a conformally invariant limit then one can show easily that the limit satisfies the restriction property . hence , the only candidate for the limit ( assuming a conformally invariant scaling limit ) is @xmath163 . the conjectures for critical exponents for saw can be interpreted in terms of rigorous properties of @xmath163 ( see @xcite ) . for example , the hausdorff dimension of @xmath163 paths is @xmath0 @xcite ; this gives strong evidence that the limit of saws should give paths of dimension @xmath0 . monte carlo simulations @xcite support the conjecture that the limit of saw is @xmath163 . aa r. bass , _ probabilistic techniques in analysis _ , springer - verlag , 1995 . v. beffara , hausdorff dimensions for @xmath99 , preprint . k. burdzy & g. lawler , non - intersection exponents for random walk and brownian motion . part ii : estimates and applications to a random fractal , _ ann . probab _ , 18 ( 1990 ) , 9811009 . j. cardy , critical percolation in finite geometries , _ j. phys . a _ , 25 ( 1992 ) , l201l206 . b. duplantier & k .- h . kwon , conformal invariance and intersections of random walks , _ phys . _ , 61 , 25142517 . b. duplantier , random walks and quantum gravity in two dimensions , _ phys . _ , 81 ( 1998 ) , 54895492 . s. fomin , loop - erased walks and total positivity , _ trans . _ , 353 ( 2001 ) , 35633583 . t. kennedy , monte carlo tests of sle predictions for the 2d self - avoiding walk , _ phys . _ 88 ( 2002 ) , 130601 r. kenyon , the asymptotic determinant of the discrete laplacian , _ acta math . _ , 185 ( 2000 ) , 239286 . r. langlands , p. pouliot , & y. saint - aubin , conformal invariance in two - dimensional percolation , _ bull . _ 90 ( 1994 ) , 161 . g. lawler , hausdorff dimension of cut points for brownian motion , _ electronic j. probab . _ , 1 ( 1996 ) , paper no . g. lawler , the dimension of the frontier of planar brownian motion , _ electronic comm . _ , 1 ( 1996 ) , paper no . 5 . g. lawler , geometric and fractal properties of brownian motion and random walk paths in two and three dimensions , in _ random walks , budapest 1998 _ , bolyai mathematical studies , 9 ( 1999 ) , 210258 . g. lawler , loop - erased random walk , _ perplexing problems in probability _ , birkhuser ( 1999 ) , 197217 . g. lawler , an introduction to the stochastic loewner evolution , preprint . g. lawler & e. puckette , the intersection exponent for simple random walk , _ combinatorics , probab . , and computing _ , 9 ( 2000 ) , 441464 . g. lawler , o. schramm , & w. werner , values of brownian intersection exponents i : half - plane exponents , _ acta . _ , 187 ( 2001 ) , 237273 . g. lawler , o. schramm , & w. werner , values of brownian intersection exponents ii : plane exponents , _ acta . _ , 187 ( 2001 ) , 275308 . g. lawler , o. schramm , & w. werner , values of brownian intersection exponents iii : two - sided exponents , _ ann . henri poincar _ , 38 ( 2002 ) , 109123 . g. lawler , o. schramm , & w. werner , analyticity of intersection exponents for planar brownian motion , _ acta . _ , to appear . g. lawler , o. schramm , & w. werner , one arm exponent for critical 2d percolation , _ electronic j. probab . _ , 7 ( 2002 ) , paper no . g. lawler , o. schramm , & w. werner , conformal invariance of planar loop - erased random walk and uniform spanning trees , preprint . g. lawler , o. schramm , & w. werner , on the scaling limit of planar self - avoiding walk , preprint . g. lawler , o. schramm , & w. werner , conformal restriction properties : the chordal case , in preparation . g. lawler & w. werner , intersection exponents for planar brownian motion , _ annals of probab . _ , 27 ( 1999 ) , 16011642 . g. lawler & w. werner , universality for conformally invariant intersection exponents , j. european math . soc . 2 ( 2000 ) , 291328 . b. mandelbrot , _ the fractal geometry of nature _ , freeman , 1982 . s. rohde & o. schramm , basic properties of sle , preprint . o. schramm , scaling limits of loop - erased random walks and uniform spanning trees , _ israel j. math _ , 118 ( 2001 ) , 221288 . o. schramm , a percolation formula , _ electronic comm . _ , 8 ( 2001 ) , paper no . 12 . s. smirnov , critical percolation in the plane : conformal invariance , cardy s formula , scaling limits , _ c. r. acad . _ , 333 ( 2001 ) , 239244 . s. smirnov & w. werner , critical exponents for two - dimensional percolation , _ math . _ , to appear . w. werner , critical exponents , conformal invariance and brownian motion , _ proceedings of the 3rd europ . congress math . math 202 ( 2001 ) , 87103 . d. wilson , generating random spanning trees more quickly than the cover time , _ proceedings of the twenty - eighth annual acm symposium on the theory of computing _ , acm ( 1996 ) , 296303 .	this paper describes joint work with oded schramm and wendelin werner establishing the values of the planar brownian intersection exponents from which one derives the hausdorff dimension of certain exceptional sets of planar brownian motion . in particular , we proof a conjecture of mandelbrot that the dimension of the frontier is @xmath0 . the proof uses a universality principle for conformally invariant measures and a new process , the stochastic loewner evolution ( @xmath1 ) , introduced by schramm . these ideas can be used to study other planar lattice models from statistical physics at criticality . i discuss applications to critical percolation on the triangular lattice , loop - erased random walk , and self - avoiding walk . 4.5 mm * 2000 mathematics subject classification : * 60j65 , 60k35 . * keywords and phrases : * brownian motion , critical exponents , conformal invariance , stochastic loewner evolution .
linteresse per le lezioni del corso di fisica teorica , svolte da ettore majorana alluniversit di napoli nel 1938 , stato recentemente ravvivato dal ritrovamento del documento moreno @xcite , una trascrizione fedele da parte di eugenio moreno degli appunti delle lezioni preparati dallo stesso majorana , comprensiva di alcune lezioni precedentemente ignote e i cui originali sembrano essere andati dispersi . gli appunti originali sono conservati a pisa , e furono riprodotti anastaticamente alcuni anni or sono @xcite , includendovi anche alcuni fogli che si voluto interpretare @xcite @xcite come gli appunti preliminari per la lezione successiva alla ventunesima , che majorana avrebbe svolto se non fosse scomparso . sebbene il carattere `` preliminare '' , ossia di appunti di uso personale da non consegnare direttamente agli studenti , sia ampiamente attestato dalle inusuali numerose cancellature che vi sono presenti , unanalisi attenta dei contenuti ivi svolti non sembra indicare che tali appunti siano stati redatti per il corso di fisica teorica n , tantomeno , preparati per la lezione imediatamente successiva alla ventunesima . infatti , come si pu constatare facilmente , tali appunti svolgono argomenti avanzati e niente affatto collegati a quanto trattato da majorana nelle sue ultime tre lezioni , in cui egli aveva appena introdotto le basi della meccanica ondulatoria e della sua interpretazione statistica . qui si presuppongono , invece , conoscenze che il docente non aveva certamente ancora fornito nel suo corso e si sviluppano delle applicazioni particolari che non possono riguardarsi come attuazioni immediate del nuovo formalismo da poco avviato ( come la teoria del legame molecolare ) . + daltra parte , lo stile adottato da majorana completamente diverso da quello impiegato nelle altre lezioni rivolte agli studenti , in cui lespressione matematica dei concetti fisici ha un ruolo rilevante , e la sua forma discorsiva si avvicina generalmente a quella adoperata nella prolusione al corso ( rivolta principalmente ai docenti delluniversit di napoli @xcite ) . in particolare sebbene negli appunti considerati non siano stati svolti tutti gli argomenti ( ovvero : le applicazioni ) riportati nella lista iniziale , la frase conclusiva di majorana alquanto perentoria e veramente tipica di una conferenza generale : `` la meccanica quantistica apre la strada per lunificazione logica di tutte le scienze che hanno per oggetto comune il mondo inorganico '' . sebbene non vi siano attualmente prove documentate dirette , e in ogni caso la questione presenti poco interesse per quanto si vuole discutere in questo luogo , per concretezza riterremo qui che tali appunti manoscritti siano stati preparati per un seminario o una conferenza generale . se tale ipotesi trovasse , poi , ulteriore credito , ci si potrebbe anche spingere a presumere che tale conferenza sia stat verosimilmente sollecitata dal direttore dellistituto di fisica delluniversit di napoli , antonio carrelli , che in quel periodo lavorava insieme con i suoi collaboratori a questioni sperimentali di fisica molecolare @xcite , e che quindi il nucleo centrale della conferenza fosse , come effettivamente appare , linterpretazione teorica del legame molecolare nella meccanica quantistica . una ipotetica data per lo svolgimento di tale conferenza potrebbe , allora , essere collocata nella seconda met di gennaio del 1938 , ossia poco dopo larrivo di majorana a napoli , e comunque non lontano dal 3 febbraio , quando egli svolse la lezione n.9 , dove affront il problema dello spettro degli atomi con due elettroni di valenza ( da un punto di vista fenomenologico ) e in cui si pu riscontrare una discussione un po pi dettagliata delle forze di risonanza ( in ambito atomico piuttosto che molecolare ; ma nel manoscritto in esame si rimanda anche a quello ) . + lattenzione del lettore , tuttavia , non vuole essere qui rivolto allaspetto _ storico _ , che pure pu avere una notevole importanza , trattandosi di un lavoro scritto comunque poco tempo prima della scomparsa dellautore , bens al contenuto _ _ del testo . sebbene , come accennato sopra , largomento principale della dissertazione debba considerarsi lapplicazione della meccanica quantistica alla teoria del legame molecolare , linteresse scientifico di essa maggiormente incentrato sulla interpretazione che majorana offre di alcuni fondamenti della ( nuova ) teoria dei quanti ( il concetto di stato quantico ) e dellapplicazione diretta di essa a un caso particolare ( appunto il legame molecolare ) . una lettura accorta del manoscritto , infatti , non solo tradisce una particolare profondit di ingegno dellautore nel trattare un argomento cardine della nuova meccanica ma , tenendo presente che esso fu scritto nel 1938 , rivela anche un notevole anticipo ( di almeno una decina di anni ) nelluso che ne viene fatto , cosa daltronde quasi usuale nel caso di majorana ( basta citare , come solo esempio , il caso dellatomo di thomas - fermi @xcite ; si veda , comunque , anche @xcite ) . + tale circostanze fu gi intuita alcuni anni or sono da n. cabibbo @xcite , il quale vide nella presunta futura lezione di majorana una vaga e approssimata anticipazione dellidea sottostante alla interpretazione di feynman della meccanica quantistica in termini degli integrali sui cammini ( _ path integral approach _ ) . uno studio pi analitico , condotta sulledizione critica dello scritto di majorana ( non disponibile a quel tempo , ma riportato alla fine del presente articolo per la prima volta ) , rivela invece delle interessanti sorprese , sulle quali si fisser qui lattenzione . sebbene la particolare semplicit e chiarezza espressiva del testo ( molto comune nelle opere di majorana @xcite ) possa far apparire come superflui ulteriori commenti , per un suo accurato inquadramento verr nel prossimo paragrafo premessa una breve discussione delle idee fondamentali della teoria degli integrali sui cammini , a cui seguir una semplice esposizione in cui solo si evidenzieranno i passaggi cruciali a tale riguardo nello scritto di majorana . il postulato generale su cui si basa la meccanica quantistica stabilisce , come noto , che lo `` stato '' di un certo sistema fisico possa rappresentarsi mediante una grandezza complessa @xmath0 , considerata come vettore ( normalizzato ) di un opportuno spazio di hilbert associato al sistema fisico , che contiene tutte le possibili informazioni sul sistema @xcite . levoluzione temporale del vettore di stato governata dallequazione di schrdinger , che pu scriversi sotto la forma generale [ e1 ] i = h , dove @xmath1 loperatore hamiltoniano del sistema considerato . lo stato iniziale al tempo @xmath2 specificato dalla scelta di questo tra i possibili autostati di un insieme completo di operatori che commutano tra di loro e con lhamiltoniana , mentre la stessa hamiltoniana @xmath1 , tramite lequazione ( [ e1 ] ) , determina lo stato del sistema a un istante successivo @xmath3 . levoluzione dinamica del sistema allora completamente determinata allorquando si calcoli lampiezza di transizione dallo stato @xmath4 al tempo @xmath2 allo stato @xmath5 al tempo @xmath3 . + come ci si pu facilmente convincere , dunque , la descrizione usuale della meccanica quantistica di un sistema fisico fortemente incentrata sul ruolo dellhamiltoniana @xmath1 e , come conseguenza , la variabile temporale gioca un ruolo centrale in tale descrizione . una siffatta dissimmetria tra le variabili spaziali e temporale , ovviamente , insoddisfacente alla luce dei postulati della teoria della relativit , e di ci se ne rese conto per primo dirac nel 1932 @xcite , il quale avanz lidea di riformulare lintera meccanica quantistica in termini lagrangiani piuttosto che hamiltoniani . + il punto di partenza nella riflessione di dirac quello di sfruttare unanalogia , a livello quantistico , con la funzione principale di hamilton nella meccanica classica @xcite . in base a ci , lampiezza di transizione per passare da uno stato individuato dalla configurazione spaziale @xmath6 al tempo @xmath7 ad uno stato individuato dalla configurazione spaziale @xmath8 al tempo @xmath9 scritta come:[e2]q_b \| q_a ~^ s = ^ _ t_a^t_b l t dove @xmath10 la lagrangiana del sistema e @xmath11 $ ] il funzionale dazione . la relazione precedente , tuttavia , non pu essere considerata unuguaglianza , almeno fintantoch lintervallo temporale fra @xmath7 e @xmath9 resta finito , in quanto condurrebbe a risultati non corretti ( e di ci era ben conscio lo stesso dirac , che nel suo articolo produsse tutta una serie di assunzioni , senza per motivate giustificazioni per esse ) . infatti , supponendo di dividere la regione dintegrazione in ( [ e2 ] ) in @xmath12 intervalli , @xmath13 , lampiezza di transizione si potrebbe scrivere come un prodotto di termini , [ e3 ] q_b \| q_a = q_b \| q_n-1 q_n-1 \| q_n-2 q_2 \| q_1 q_1 \| q_a , mentre ben noto , usando le relazioni di completezza , che la formula corretta contiene le integrazioni sulle regioni intermedie : q_b \| q_a & = & q_b \| q_n-1 \| q_n-1 q_n-1 \| q_1 \| q_1 q_1 \| q_a + & = & q_1 q_2 q_n-1 q_b \| q_n-1 q_2 \| q_1 q_1 \| q_a . [ e4 ] lintuizione di feynman @xcite , palesata circa dieci anni dopo larticolo di dirac , fu allora quella di ritenere che la ( [ e2 ] ) valesse come uguaglianza , a meno di una costante moltiplicativa @xmath14 , solo per transizioni tra stati separati da un intervallo di tempo _ infinitesimo_. in tal caso , dunque , utilizzando la formula corretta ( [ e4 ] ) , si ottiene la ben nota espressione di feynman per lampiezza di transizione fra due stato generici : [ e5 ] q_b \| q_a = _ n + t_b - t_a finito a^n q_1 q_2 q_n-1 ^i s/ q ^i s/ . il significato della formula precedente pu comprendersi nel modo seguente . nelle integrazioni presenti , che si indicato genericamente con @xmath15 , gli estremi @xmath7 e @xmath9 dellintervallo di integrazione sono tenuti fissi , mentre si integra sullintero spazio rispetto rispetto ai punti intermedi . poich ogni configurazione spaziale @xmath16 dei punti intermedi corrisponde ad una data traiettoria dinamica che unisce il punto iniziale @xmath7 con quello finale @xmath9 , lintegrazione su tutte queste configurazioni equivale allora a sommare su _ tutti _ i possibili cammini che collegano il punto iniziale con quello finale . in altre parole la formula di feynman con gli integrali sui cammini evidenzia che lampiezza di transizione tra uno stato iniziale e uno stato finale pu essere espresso come una somma su tutti i cammini , con estremi fissati , del fattore @xmath17/\hbar}$ ] . tale risultato , da un lato , non affatto sorprendente se si pensa al fatto che , nella meccanica quantistica , se un dato processo pu avvenire secondo modalit diverse , lampiezza di transizione data dalla somma delle ampiezze individuali corrispondenti a _ tutti _ i possibili modi in cui il processo pu avvenire . ci risulta molto chiaro , ad esempio , dal classico esperimento con un fascio di elettroni che attraversa uno schermo con una doppia fenditura . la struttura interferenziale che si osserva dietro lo schermo ad una certa distanza da esso , che sembrerebbe indicare che uno stesso elettrone sia passato attraverso entrambe le fenditure , infatti spiegabile ammettendo che la probabilit che lelettrone passi dalla sorgente allo schermo di raccolta attraverso la doppia fenditura sia ottenuto sommando su _ tutti _ i possibili percorsi dellelettrone . lapproccio on gli integrali sui cammini , dunque , contiene al suo interno i principi fondamentali della meccanica quantistica . tuttavia ci che in esso cruciale e inaspettato che la somma eseguita sul fattore di fase @xmath17/\hbar}$ ] , generato dallazione classica @xmath11 $ ] . + uno dei migliori pregi dellapproccio di feynman alla meccanica quantistica la possibilit di ottenere in modo estremamente chiaro il passaggio alla meccanica classica nel limite @xmath18 ( gli altri pregi essendo la sua versatile applicabilit alla quantizzazione dei campi nelle teorie di gauge abeliane , non abeliane e con rottura spontanea della simmetria di gauge ) . infatti , per grandi valori di @xmath5 rispetto ad @xmath19 , il fattore di fase nella ( [ e5 ] ) subisce delle grandi fluttuazioni e contribuisce quindi con termini che si mediano a zero . da un punto di vista matematico , dunque , risulta chiaro che nel limite @xmath20 il contributo dominante alla ( [ e5 ] ) viene fuori nel caso in cui il fattore di fase non cambia molto o , ci che lo stesso , quando lazione @xmath5 stazionaria . tale risultato appunto ci che si riscontra nel caso della meccanica classica , dove le traiettorie dinamiche classiche si ottengono dal principio di minima azione . questa circostanza fu notata per primo da dirac @xcite , il quale si rese ben conto del ruolo molto importante giocato dal funzionale dazione . + linterpretazione intuitiva del passaggio al limite classico molto semplice . nello spazio @xmath21 si consideri un dato cammino che sia lontano dalla traiettoria classica @xmath22 ; poich @xmath19 piccolo l fase @xmath23 lungo tale cammino sar abbastanza grande . per ognuno di tali cammini ce ne sar uno vicino , infinitamente prossimo ad esso , dove lazione @xmath5 cambier solo per una piccola quantit , ma poich questa moltiplicata per una costante ( @xmath24 ) molto grande , la fase risultante avr un valore altrettanto grande . in media tali cammini daranno un contributo nullo nella somma ( [ e5 ] ) . vicino la traiettoria classica @xmath22 , invece , poich lazione stazionaria , passando ad un cammino infinitamente prossimo a quello classico lazione non cambier affatto , per cui i i corrispondenti contributi alla ( [ e5 ] ) si sommeranno coerentemente , e conseguentemente si otterr lapporto dominante per @xmath25 . in tale approccio , dunque , nel limite @xmath20 viene selezionata la traiettoria classica non perch essa contribuisca maggiormente allevoluzione dinamica del sistema , ma piuttosto perch vi sono cammini infinitamente prossimi ad essa che danno contributi che si sommano coerentemente . la regione di integrazione allora effettivamente molto sottile nel caso di sistemi classici , mentre diviene molto larga per sistemi quantistici . di conseguenza il concetto stesso di orbita , che nel caso classico ben definito , in situazioni quantistiche viene a perdere il suo significato , come ad esempio nel caso di un elettrone che ruota intorno al nucleo di un atomo . la discussione presentata sopra dellapproccio alla meccanica quantistica in termini di integrali sui cammini si incentrata volutamente sullaspetto matematico piuttosto che su quello fisico , in quanto stato esattamente questo il processo storico che ha portato dallidea originaria di dirac del 1932 alla formulazione di feynman negli anni 40 ( basta per questo consultare gli articoli originali in @xcite e @xcite ) . daltra parte stato proprio lo sviluppo del formalismo matematico che ha permesso , successivamente , la notevole sua interpretazione fisica accennata nel paragrafo precedente . + ritornando allo scritto di majorana qui in esame , si scorge subito che _ nulla _ dellaspetto matematico del peculiare approccio alla meccanica quantistica presente in esso . tuttavia una lettura attenta rivela in maniera altrettanto evidente la presenza in esso dei fondamenti _ fisici _ della stessa . + il punto di partenza di majorana il voler ricercare una formulazione quanto pi significativa e chiara del concetto di stato quantico . e , naturalmente , nel 1938 la polemica con le concezioni della vecchia teoria dei quanti ancora aperta . + _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ secondo la teoria di heisenberg uno stato quantico corrisponde non a una soluzione stranamente privilegiata delle equazioni classiche ma piuttosto a un complesso di soluzioni differenti per le condizioni iniziali e anche per lenergia , ci che si presenta come energia esattamente definita , dello stato quantico corrispondendo a una specie di media presa sulle infinite orbite classiche raccolte in esso . gli stati quantici si presentano cos come i minimi complessi statistici di movimenti classici accessibili allosservazione . questi complessi statistici minimi non si lasciano ulteriormente dividere a causa del principio di indeterminazione , enunciato dallo stesso heisenberg , che vieta la misura esatta contemporanea della posizione e della velocit di un particella e cos la fissazione della sua orbita . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ si osservi che le `` soluzioni differenti per le condizioni iniziali '' corrispondono , nel linguaggio di feynman del 1948 , proprio ai diversi cammini di integrazione ; infatti le diverse condizioni iniziali sono comunque sempre riferite ad uno stesso tempo iniziale ( @xmath7 ) , mentre il determinato stato quantico in oggetto corrisponde ad un fissato tempo finale ( @xmath9 ) . la questione introdotta dei `` movimenti classici '' ( la sottolineatura di majorana ) , secondo la precisazione , immediatamente seguente , che ne fornisce il principio di indeterminazione di heisenberg , pertanto palesemente connessa a quella dellestensione sufficientemente ampia della regione dintegrazione in ( [ e5 ] ) per sistemi quantistici , che viene cos ad essere intimamente legata ad un principio fisico fondamentale . + il punto cruciale della formulazione di feynman della meccanica quantistica , come si visto , il considerare _ tutti _ i possibili cammini che uniscono il punto iniziale con quello finale , e non solo quelli corrispondenti alle traiettorie classiche . nel testo di majorana , dopo aver discusso un interessante esempio relativo alloscillatore armonico , lautore rimarca : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ naturalmente la corrispondenza fra stati quantici e complessi di soluzioni classiche non che approssimata poich le equazioni della dinamica quantistica sono in generale indipendenti dalle corrispondenti equazioni classiche e indicano una reale modificazione delle leggi meccaniche oltre che un limite posto alle possibilit di osservazione ; ma essa meglio fondata della rappresentazione degli stati quantici mediante orbite quantizzate e pu essere utilmente impiegata in considerazioni di carattere qualitativo . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ e , ancora pi esplicitamente , in un luogo pi avanti , si dice che la funzione donda `` associata nella meccanica quantistica ad ogni stato possibile dellelettrone '' . tale riferimento , che superficialmente potrebbe essere interpretato nel senso usuale che la funzione donda contiene tutte le informazioni del sistema fisico , invece da considerarsi secondo laccezione di feynman , alla luce della approfondita discussione che majorana compie sul concetto di stato . + infine da evidenziare come nellanalisi di majorana giochino un ruolo fondamentale anche le propriet di simmetria del sistema fisico . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ sotto certe ipotesi che , nei problemi assai semplici che esamineremo , sono verificate si pu dire che ogni stato quantico possiede tutti i caratteri di simmetria dei vincoli del sistema . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ lattinenza con la formulazione in termini di integrali sui cammini pu riferirsi a quanto segue . nel discutere un dato sistema atomico majorana sottolinea come da uno stato quantico @xmath5 del sistema si possa passare ad un altro stato @xmath26 mediante unoperazione di simmetria . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a differenza per di ci che accade nella meccanica classica per le delle equazioni dinamiche , non pi vero in generale che @xmath26 sia diverso da @xmath5 . possiamo rendercene conto facilmente associando a @xmath26 , come si visto , un complesso di soluzioni delle equazioni classiche , poich basta allora che @xmath5 comprenda con ogni soluzione anche laltra che si deduce dalla prima in base a una propriet di simmetria nei moti del sistema perch @xmath26 risulti costituito identicamente a @xmath5 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ tale passaggio peculiarmente intrigante se si osserva che la questione della ridondanza nella misura dintegrazione per teorie che esibiscono simmetrie di gauge , e che porta ad espressioni infinite per lampiezza di transizione @xcite , fu sollevata ( e risolta ) solo molto tempo dopo il lavoro iniziale di feynman . + in conclusione , sebbene sia indubbio che nel testo di majorana non vi sia alcuna traccia indicativa del formalismo sottostante allapproccio di feynman alla meccanica quantistica in termini di integrali sui cammini ( come invece presente nellarticolo di dirac del 1933 che forse majorana conosceva ) , comunque molto interessante notare che in esso sono presenti i caratteri fisici salienti del nuovo modo di interpretare la teoria dei quanti . e tale circostanza particolarmente significativa se si ha presente il fatto che , nellesposizione storica nota , linterpretazione del formalismo venuto soltanto dopo lo sviluppo di questultimo . + nel manoscritto di majorana , poi , sono presenti anche alcune interessanti applicazioni a sistemi atomici e molecolari , in cui risultai noti sono ricavati o reinterpretati secondo il nuovo punto di vista . di questi , per , si lascer la scoperta la lettore nelle pagine che seguono , dove verr riportato interamente il testo qui discusso . il manoscritto di majorana , come pu vedersi in @xcite , presenta al suo inizio un indice numerato degli argomenti da svolgere , solo parzialmente rispettato nel testo che segue . per comodit di lettura si qui preferito suddividere lintero scritto in paragrafi secondo lindice iniziale . + @xmath27 lenergia interna di un sistema chiuso ( atomo , molecola , etc . ) suscettibile secondo la meccanica quantistica di una successione discreta di valori @xmath28 che costituiscono i cosiddetti `` autovalori '' dellenergia . a ogni determinazione dellenergia corrisponde uno `` stato quantico '' , cio uno stato in cui il sistema pu permanere indefinitamente in assenza di perturbazioni esterne . fra queste si deve in generale considerare laccoppiamento del sistema con il campo di radiazione in virt del quale esso pu perdere energia sotto forma di irraggiamento elettromagnetico passando da un livello energetico @xmath29 a un livello inferiore @xmath30 . solo quando lenergia interna ha il valore minimo @xmath31 , essa non pu diminuire ulteriormente per irradiazione ; si dice allora che il sistema nel suo `` stato fondamentale '' da cui non pu essere rimosso senza lintervento di influenze esterne sufficientemente intense , come lurto con particelle veloci o con quanti di luce di frequenza elevata . + che cosa corrisponde nella meccanica classica a uno stato quantico ? a questa domanda si deve anzitutto rispondere se si vuol giungere a una rappresentazione corretta dei risultati a cui pervenuta nel nostro campo la meccanica quantistica , senza tuttavia entrare a fondo nei complessi metodi di calcolo da questa adottati . + nella meccanica classica il movimento di un sistema composto di @xmath12 punti materiali interamente determinato quando sono note le coordinate @xmath32 di tutti punti in funzione del tempo : [ a1 ] q_i = q_i(t ) le ( [ a1 ] ) risolvono le equazioni della dinamica in cui figurano tutte le forze interne ed esterne a cui il sistema soggetto e possono sempre scegliersi in modo che per un certo istante tutte le coordinate @xmath33 e le loro derivate temporali @xmath34 abbiano valori arbitrariamente prefissati . cos la soluzione generale delle equazioni del movimento deve contenere @xmath35 costanti arbitrarie . + per sistemi di proporzioni atomiche la rappresentazione classica non pi valida e ne sono state proposte successivamente due modificazioni . la prima , che dovuta a bohr e sommerfeld e ha reso lungamente utili servizi , stata in seguito interamente abbandonata di fronte al sorgere della nuova meccanica quantistica che sola ha fornito un formalismo di estrema generalit , e pienamente confermato dallesperienza per lo studio dei fenomeni elementari . secondo la vecchia teoria di bohr - sommerfeld la meccanica classica ancora valida allinterno dellatomo cos che il movimento dellelettrone ad es . intorno al nucleo di idrogeno ancora descritto mediante una soluzione ( [ a1 ] ) delle equazioni della meccanica classica ; per se si considerano movimenti aventi carattere periodico , come appunto la rivoluzione di un elettrone intorno al nucleo , non tutte le soluzioni delle equazioni classiche sono realizzate in natura , ma solo uninfinit discreta di esse soddisfacente alle cosiddette condizioni di sommerfeld , cio a certe relazioni integrali di sapore quasi cabalistico . per esempio in ogni moto periodico in una dimensione lintegrale esteso a tutto il periodo @xmath36 : @xmath37 del doppio dellenergia cinetica deve essere multiplo intero della costante di planck ( @xmath38 ) . lunione della meccanica classica e di un principio estraneo al suo spirito , come quello delle orbite quantizzate , appare cos ibrida che non deve recare meraviglia la rovina completa della teoria avvenuta nellultimo decennio ad onta di varie prove sperimentali in suo favore che si erano credute decisive . + la nuova meccanica quantistica , dovuta precipuamente a heisenberg , sostanzialmente pi prossima alle concezioni classiche che non la vecchia . secondo la teoria di heisenberg uno stato quantico corrisponde non a una soluzione stranamente privilegiata delle equazioni classiche ma piuttosto a un complesso di soluzioni differenti per le condizioni iniziali e anche per lenergia , ci che si presenta come energia , esattamente definita , dello stato quantico corrispondendo a una specie di media presa sulle infinite orbite classiche raccolte in esso . gli stati quantici si presentano cos come i minimi complessi statistici di movimenti classici accessibili allosservazione . questi complessi statistici minimi non si lasciano ulteriormente dividere a causa del principio di indeterminazione , enunciato dallo stesso heisenberg , che vieta la misura esatta contemporanea della posizione e della velocit di un particella e cos la fissazione della sua orbita . + un oscillatore armonico di frequenza @xmath39 pu oscillare classicamente con ampiezza e fase arbitrarie , la sua energia essendo data da @xmath40 se @xmath41 la sua massa e @xmath42 lelongazione massima . secondo la meccanica quantistica i valori possibili di @xmath43 sono , come noto , @xmath44 , @xmath45 , @xmath46 ; in questo caso si pu dire che lo stato fondamentale con energia @xmath44 corrisponde grosso modo a tutte le oscillazioni classiche con energia compresa fra @xmath47 e @xmath48 , il primo stato eccitato con energia @xmath49 corrisponde alle soluzioni classiche con energia compresa fra @xmath50 e @xmath51 , e cos via . naturalmente la corrispondenza fra stati quantici e complessi di soluzioni classiche non che approssimata poich le equazioni della dinamica quantistica sono in generale indipendenti dalle corrispondenti equazioni classiche e indicano una reale modificazione delle leggi meccaniche oltre che un limite posto alle possibilit di osservazione ; ma essa meglio fondata della rappresentazione degli stati quantici mediante orbite quantizzate e pu essere utilmente impiegata in considerazioni di carattere qualitativo . @xmath27 uno studio particolare meritano i sistemi aventi qualche carattere di simmetria . per essi da una soluzione particolare delle equazioni classiche del movimento @xmath52 se ne deducono in generale altre differenti @xmath53 in base a sole considerazioni di simmetria . per esempio se il sistema contiene due o pi elettroni o in genere due o pi particelle uguali si pu , scambiando le coordinate di due particelle passare da una soluzione a unaltra che sar in generale differente . analogamente se consideriamo un elettrone mobile nel campo di due nuclei o atomi uguali ( nella figura @xmath14 e @xmath54 ) si pu , riflettendo nel centro @xmath55 del segmento @xmath56 , dedurre da unorbita possibile @xmath52 percorsa con determinata legge oraria intorno ad @xmath14 , unorbita @xmath53 descritta dallelettrone intorno al nucleo o atomo @xmath54 . = 6.5truecm le operazioni di scambio di due particelle uguali , riflessioni in un punto , o altre corrispondenti a caratteri qualsiasi di simmetria , conservano il loro significato nella meccanica quantistica . cos possibile dedurre da uno stato quantico @xmath5 un altro stato @xmath26 , , se nei due esempi citati si scambiano due particelle uguali e rispettivamente si riflette il sistema nel punto @xmath55 . a differenza per di ci che accade nella meccanica classica per le delle equazioni dinamiche , non pi vero in generale che @xmath26 sia diverso da @xmath5 . possiamo rendercene conto facilmente associando a @xmath26 , come si visto , un complesso di soluzioni delle equazioni classiche , poich basta allora che @xmath5 comprenda con ogni soluzione anche laltra che si deduce dalla prima in base a una propriet di simmetria nei moti del sistema perch @xmath26 risulti costituito identicamente a @xmath5 . + in alcuni casi se il sistema presenta caratteri di simmetria sufficientemente complessi in realt possibile ottenere per simmetria da uno stato quantico altri stati diversi ma aventi la stessa energia . si dice allora che il sistema , ha cio parecchi stati di uguale energia , in virt appunto dei suoi caratteri di simmetria . lo studio dei sistemi degeneri e delle condizioni in cui pu aversi degenerazione ci porterebbe troppo lontano ed del resto difficile a eseguirsi in base a sole analogie classiche . lo lasceremo peci interamente da parte confinando la nostra attenzione su problemi in cui non si presenta degenerazione . questa condizione sempre soddisfatta se la simmetria del sistema meccanico consente solo una trasformazione cos semplice che il suo quadrato , cio la sua ripetizione due volte consecutivamente , si riduce a una trasformazione identica . per esempio , riflettendo due volte di seguito un sistema di punti materiali rispetto a un piano , a una retta o a un punto ritroviamo necessariamente la stessa disposizione iniziale ; analogamente scambiando due volte due particelle uguali il sistema rimane inalterato . in tutti questi casi si presentano solo stati quantici semplici , cio a ogni valore possibile dellenergia appartiene un solo stato quantico . segue che tutti gli stati quantici di un sistema contenente due particelle uguali sono simmetrici nelle due particelle , restando inalterati per il loro scambio . e cos gli stati di un elettrone circolante intorno a due nuclei uguali @xmath14 e @xmath54 sono simmetrici rispetto al punto medio @xmath55 di @xmath56 , o restano inalterati per riflessione in @xmath55 , e analogamente in altri casi simili . sotto certe ipotesi che , nei problemi assai semplici che esamineremo , sono verificate si pu dire che ogni stato quantico possiede tutti i caratteri di simmetria dei vincoli del sistema . consideriamo ora un elettrone mobile nel campo di due nuclei di idrogeno o protoni . il sistema dei due protoni e dellelettrone ha una carica netta risultante @xmath57 , e costituisce la pi semplice molecola realizzabile , cio la molecola di idrogeno ionizzata positivamente . in tale sistema sono mobili cos i protoni come lelettrone , ma a causa della grande differenza di massa fra i primi e il secondo ( rapporto delle masse 1840:1 ) la velocit media dei protoni molto inferiore a quella dellelettrone , e il movimento di questultimo si pu studiare con grande approssimazione supponendo i protoni fermi a una certa distanza reciproca . questa distanza determinata da ragioni di stabilit in modo che sia minima lenergia totale della molecola , che risulta in prima approssimazione dalla somma dellenergia potenziale mutua dei due protoni e dellenergia dellelettrone mobile nel campo dei primi due , ed diversa per i diversi stati quantici dellelettrone . + lenergia potenziale mutua dei protoni data da @xmath58 se @xmath59 la loro distanza , mentre lenergia di legame dellelettrone nel suo stato fondamentale una funzione negativa @xmath60 di @xmath59 che non ha unespressione analitica semplice ma pu essere calcolata dalla meccanica quantistica con approssimazione grande quanto si desideri . la distanza @xmath61 di equilibrio cos determinata dalla condizione che sia minima lenergia totale @xmath62 la curva @xmath63 ha un andamento del tipo indicato in figura , se si assume come zero dellenergia quella corrispondente alla molecola dissociata in un atomo neutro di idrogeno e un atomo ionizzato a distanza infinita . la distanza di equilibrio stata = 3.5 cm calcolata teoricamente da burrau che ha trovato @xmath64 cm e per lenergia corrispondente @xmath65 volt - elettrone . questi risultati sono stati entrambi pienamente confermati dallosservazione a cui sono indirettamente accessibili attraverso lo studio dello spettro emesso dalla molecola neutra o ionizzata . + qual lorigine della forza @xmath66 che tende ad avvicinare i due nuclei di idrogeno quando essi si trovano a una distanza maggiore di @xmath61 ? la risposta che la meccanica quantistica d a tale questione sorprendente poich sembra indicare che accanto a certe forze di polarizzazione prevedibili in base alla meccanica classica hanno una parte preponderante forze di un tipo assolutamente nuovo , le cosiddette forze di risonanza . + supponiamo la distanza @xmath59 grande rispetto al raggio dellatomo neutro di idrogeno ( @xmath67 ) . lelettrone si trova allora sotto lazione delluno o dellaltro dei due protoni intorno a ognuno dei quali pu descrivere classicamente orbite chiuse . il sistema formato dallelettrone e dal nucleo intorno a cui si muove costituisce un atomo neutro di idrogeno , cos che la nostra molecola risulta essenzialmente composta da un atomo neutro e da un protone a una certa distanza dal primo . latomo neutro di idrogeno nel suo stato fondamentale ha una distribuzione di carica a simmetria sferica , ci che classicamente significa che tutte le orientazioni dellorbita elettronica sono egualmente possibili , la densit di carica negativa decrescendo esponenzialmente con la distanza in modo che il raggio dellatomo si presenta praticamente come finito ; segue che nessun campo elettrico generato allesterno da un atomo neutro di idrogeno e cos nessuna azione pu essere esercitata su un protone posto a una distanza @xmath59 grande rispetto alle dimensioni dellatomo . in realt per latomo neutro si polarizza sotto lazione del protone esterno e acquista un momento elettrico nella direzione protone - atomo neutro , e dallinterazione di questo momento elettrico con il campo non uniforme generato dal protone sorge una forza attrattiva che tende ad associare atomo e ione in un sistema molecolare . + le forze di polarizzazione facilmente prevedibili con considerazioni classiche posso dare origine da sole a composti molecolari , le cosiddette molecole di polarizzazione , che sono per caratterizzate da unestrema labilit . complessi molto stabili possono aversi solo se altre forze si aggiungono a quelle di polarizzazione . nelle molecole polari , composte di due ioni di segno differente , queste forze sono date essenzialmente dallattrazione elettrostatica fra gli ioni ; ad es . la molecola @xmath68 tenuta insieme essenzialmente dallattrazione mutua fra lo ione positivo @xmath69 e lo ione negativo @xmath70 . ma in molecole composte di due atomi neutri , o di un atomo neutro e un atomo ionizzato , come nel caso dello ione molecolare @xmath71 , laffinit chimica essenzialmente condizionata dal fenomeno di risonanza nel significato , senza analogia nella meccanica classica , che questa parola ha assunto nella nuova meccanica . + quando si studia dal punto di vista della meccanica quantistica il movimento dellelettrone nel campo dei due protoni supposti fissi a una distanza mutua @xmath59 molto grande si possono in prima approssimazione determinarne i livelli energetici supponendo che lelettrone debba muoversi intorno al protone @xmath14 ( o @xmath54 ) e prescindendo dallesistenza dellaltro protone @xmath54 ( o @xmath14 ) che esercita unazione perturbativa debole a causa della sua distanza . si ottiene cos per il pi basso autovalore dellenergia @xmath31 uno stato @xmath5 che corrisponde alla formazione di un atomo neutro nello stato fondamentale fra lelettrone e il nucleo @xmath14 e uno stato @xmath26 che corrisponde a un atomo neutro formato fra lelettrone e il nucleo @xmath54 . se ora si tiene conto della perturbazione che in ognuno dei due casi esercitata sullatomo neutro dallo ione positivo noi troveremo ancora , finch la perturbazione piccola , in luogo di due autovalori uguali @xmath31 due autovalori @xmath72 e @xmath73 poco differenti perch entrambi prossimi a @xmath31 ; ma gli stati quantici ad essi corrispondenti , e siano essi @xmath74 e @xmath75 , non sono prossimi separatamente a @xmath5 e @xmath26 , perch essendo il campo di potenziale in cui lelettrone si muove simmetrico rispetto al punto medio di @xmath56 , la stessa simmetria devono presentare , per quanto si detto prima , gli stati effettivi @xmath74 e @xmath75 dellelettrone , mentre non la presentano separatamente @xmath5 e @xmath26 . + secondo la rappresentazione modellistica degli stati quantici che abbiamo descritto pi sopra , @xmath5 composto da un complesso di orbite elettroniche circolanti intorno ad @xmath14 , @xmath26 analogamente da un complesso di orbite intorno a @xmath54 , mentre gli stati quantici veri del sistema @xmath74 e @xmath75 abbracciano ognuno , in prima approssimazione per @xmath59 molto grande , met delle orbite costituenti @xmath5 e met di quelle che costituiscono @xmath26 . il calcolo prova che per distanze nucleari sufficientemente grandi il valor medio degli autovalori perturbati @xmath72 e @xmath73 coincide sensibilmente con lunico valore imperturbato @xmath31 , mentre la loro differenza apprezzabile e ha importanza decisiva in questo e in infiniti altri casi analoghi per lo studio delle reazioni chimiche . si pu dunque supporre @xmath76 ma @xmath77 e sar allora @xmath74 lo stato fondamentale dellelettrone mentre @xmath75 apparir come stato eccitato con energia leggermente superiore . + lelettrone nello stato @xmath74 , come pure nello stato @xmath75 , passa met del suo tempo intorno al nucleo @xmath14 e laltra met intorno al nucleo @xmath54 . si pu anche stimare la frequenza media del passaggio periodico dellelettrone da @xmath14 a @xmath54 e viceversa , e cos dallo scambio dello stato , neutro o ionizzato , dei due atomi , e si trova @xmath78 essendo @xmath79 la costante di planck . per grandi valori di @xmath59 , @xmath80 decresce secondo una curva di tipo esponenziale e cos la frequenza di scambio tende rapidamente a zero , e questo significa che lelettrone posto inizialmente intorno ad @xmath14 vi rimane per un tempo sempre pi lungo , ci che ben comprensibile da un punto di vista classico . + se lelettrone si trova nello stato @xmath74 , cio nel suo stato fondamentale , la sua energia ( @xmath72 ) inferiore a quella ( @xmath73 ) che esso avrebbe senza laccennato effetto di scambio fra lufficio dei nuclei @xmath14 e @xmath54 . questa circostanza d origine a un nuovo tipo di forze attrattive fra i nuclei che si aggiungono alle forze di polarizzazione considerate prima e sono precisamente la causa determinante dellunione molecolare . + le forze di risonanza non hanno come abbiamo detto analogia nella meccanica classica . in realt fin dove pu condurci lanalogia che ci ha permesso di associare a uno stato quantico un complesso statistico di movimenti classici , i due stati @xmath74 e @xmath75 in cui pure le forze di risonanza hanno segno opposto risultano identicamente costituiti , ognuno dalla met di entrambi gli originari stati imperturbati @xmath5 e @xmath26 . ma questo vero soltanto in una certa approssimazione , cio nellapprossimazione stessa in cui sono trascurabili le forze di risonanza . per un calcolo esatto che tenga conto delle forze di risonanza si deve usare necessariamente la meccanica quantistica e si trova allora una differenza nella struttura dei due stati quantici che si manifesta soprattutto nella regione intermedia fra @xmath14 e @xmath54 attraverso la quale con un meccanismo non descrivibile secondo la meccanica classica ha luogo il passaggio periodico dellelettrone da un atomo allaltro . questa differenza qualitativa ha un carattere assolutamente formale e possiamo enunciarla solo attraverso lintroduzione della funzione donda @xmath81 che , come noto , associata nella meccanica quantistica ad ogni stato possibile dellelettrone . il modulo del quadrato di @xmath0 , che pu anche essere una grandezza complessa , d la probabilit che lelettrone si trovi nellunit di volume intorno a un punto generico @xmath82 . la funzione donda @xmath0 deve allora soddisfare a unequazione differenziale lineare e cos sempre lecito moltiplicare @xmath0 in tutti i punti per uno stesso numero reale o complesso di modulo @xmath83 , questultima limitazione essendo richiesta dalla condizione di normalizzazione @xmath84 necessaria per laccennata interpretazione fisica di @xmath85 . la moltiplicazione di @xmath0 per una costante di modulo @xmath83 lascia inalterata la distribuzione spaziale della carica elettronica e non ha in generale alcun significato fisico . noi definiremo ora formalmente la riflessione di uno stato quantico nel punto @xmath55 medio dei due nuclei @xmath14 e @xmath54 direttamente sulla funzione donda @xmath0 , ponendo @xmath86 in un sistema di coordinate aventi lorigine in @xmath55 . se @xmath0 deve rappresentare uno stato quantico simmetrico e quindi invariante per riflessione in @xmath55 , la funzione donda riflessa @xmath87 deve avere lo stesso significato fisico di @xmath0 e cos differire da @xmath0 , per quanto si detto , per un fattore costante , reale o complesso di modulo @xmath83 . questo fattore costante deve inoltre valere @xmath88 , poich il suo quadrato deve dare lunit come prova il fatto riflettendo ancora @xmath87 nel punto @xmath55 si ottiene di nuovo la funzione donda iniziale @xmath0 . + per tutti gli stati del nostro sistema si dovr dunque avere : @xmath89 dovendosi scegliere il segno @xmath90 per una parte di essi e il segno @xmath91 per gli altri . la differenza formale fra gli stati @xmath74 e @xmath75 considerati pi sopra consiste precisamente in questo che nellequazione precedente per @xmath74 valido il segno superiore ma per @xmath75 quello inferiore . la simmetria rispetto a un punto , e cos in genere ogni propriet di simmetria , determina una distinzione formale degli stati del sistema in due o pi categorie , una propriet importante di questa distinzione essendo che non possono provocarsi transizioni fra stati appartenenti a categorie diverse mediante perturbazioni esterne che rispettino le simmetrie esistenti nei vincoli del sistema . cos in sistemi contenenti due elettroni si hanno due categorie di stati non combinabili distinti secondo che la funzione donda , che ora dipende dalle coordinate di entrambi gli elettroni , resta inalterata o cambia segno scambiando le due particelle uguali . nel caso speciale dellatomo di elio questo d luogo alla ben nota apparenza spettroscopica di due elementi distinti : parelio e ortoelio . + la teoria dellaffinit chimica fra atomo neutro e atomo ionizzato di idrogeno che abbiamo considerato fin qui pu essere estesa per lo studio della molecola neutra di idrogeno e pi in generale di tutte le molecole risultanti da due atomi neutri uguali . per la molecola neutra di idrogeno dovremo considerare in luogo di un solo elettrone mobile intorno a due protoni fissi , elettroni mobili nello stesso campo prescindendo in una prima approssimazione dalla loro mutua repulsione . la stabilit della molecola allora comprensibile assumendo che ognuno dei due elettroni cada nello stato @xmath74 a cui corrispondono forze di risonanza attrattive . si pu dire con f. hund che la molecola di idrogeno tenuta insieme da due elettroni `` leganti '' . in realt linterazione degli elettroni cos forte da lasciare solo una giustifica qualitativa alla teoria schematica di hund , ma nulla vieta in principio di prevedere con esattezza tutte le propriet della molecola di idrogeno ... risolvendo con sufficiente precisione le equazioni poste dalla meccanica quantistica . cos si potuto , con metodi matematici appropriati , determinare effettivamente in base a pure considerazioni teoriche laffinit chimica fra due atomi neutri di idrogeno e il valore teorico si accorda con quello sperimentale , entro i limiti di approssimazione in cui , per ragioni pratiche , si voluto condurre il calcolo . + per molecole diverse da quella di idrogeno la teoria dellaffinit chimica notevolmente pi complicata sia per il maggior numero di elettroni che occorre considerare , sia a causa del principio di pauli che vieta la contemporanea presenza di pi di due elettroni nello stesso stato ; ma le diverse teorie dellaffinit chimica che sono state proposte negli ultimi anni e di cui ognuna ha un campo pi o meno largo di validit , consistono in sostanza nella ricerca di metodi di calcolo approssimati per la risoluzione di un problema matematico che in s esattamente determinato , non gi nellenunciazione di nuovi principi fisici . + cos possibile ricondurre la teoria delle saturazioni delle valenze ai pi generali principi della fisica . la meccanica quantistica apre la strada per lunificazione logica di tutte le scienze che hanno per oggetto comune il mondo inorganico . sono molto grato al dr . alberto de gregorio per le discussioni e i preziosi suggerimenti offerti , nonch per una attenta lettura di una versione preliminare del presente articolo . ringrazio inoltre il prof . erasmo recami per la fattiva collaborazione e per numerose discussioni . s. esposito , _ il corso di fisica teorica di ettore majorana : il ritrovamento del documento moreno _ , il nuovo saggiatore * 21 * n.1 - 2 ( 2005 ) 21 . e. majorana , _ lezioni alluniversit di napoli _ ( a cura di b. preziosi ) ( bibliopolis , napoli , 1987 ) . e. di grezia e s. esposito , _ fermi , majorana and the statistical model of atoms _ , found . * 34 * ( 2004 ) 1431 . + si veda anche s. esposito , _ majorana solution of the thomas - fermi equation _ , am . j. phys . * 70 * ( 2002 ) 852 . si pu utilmente consultare , ad esempio , p. caldirola , r. cirelli e g.m . prosperi , _ introduzione alla fisica teorica _ , ( utet , torino , 1982 ) . e. majorana , _ teoria relativistica di particelle con momento intrinseco arbitrario _ , il nuovo cimento * 9 * ( 1932 ) 335 . e. majorana , _ teoria simmetrica dellelettrone e del positrone _ , il nuovo cimento * 14 * ( 1937 ) 171 . h. goldstein , _ meccanica classica _ ( zanichelli , bologna , 1986 ) . r. p. feynman , _ the principle of least action in quantum mechanics _ ( princeton university , ann arbor , 1942 ; university microfilms publication no . + r. p. feynman , _ space - time approach to non - relativistic quantum mechanics _ , rev . phys . * 20 * ( 1948 ) . 367 r. p. feynman , _ the quantum theory of gravitation _ , acta phys . pol . * 24 * ( 1963 ) 697 . + l. d. faddeev e v. n. popov , _ feynman diagrams for the yang - mills field _ , phys . lett . * b25 * ( 1967 ) 29 .	si riporta per la prima volta il testo critico completo di un manoscritto di ettore majorana , conservato insieme agli appunti per le lezioni tenute alluniversit di napoli e redatto probabilmente per un seminario svolto in quella universit . alcuni passaggi del testo rivelano , in forma latente , una interpretazione fisica della meccanica quantistica che anticipa di molti anni lapproccio di feynman in termini di integrali sui cammini , indipendentemente dalla formulazione matematica sottostante ad esso .
since the location of our solar system is within the galactic disk , studying the galactic structure of our milky way is always a challenging problem . therefore , we can not derive such comprehensive and intuitive pictures of our disk as extragalactic studies are able to do for other spiral galaxies ( e.g. , @xcite ) . nevertheless , based on a diverse set of studies over all wavelengths , in the last few decades we have derived a reasonably comprehensive picture of our galactic spiral structure ( for recent work , see , e.g. , @xcite ) . the galactic plane has been observed in the optical / near-/mid - infrared bands ( e.g. , @xcite ) as well as at longer wavelengths , e.g. , in co or cm continuum emission ( e.g. , @xcite ) . however , until the arrival of the two ( sub)mm galactic plane surveys atlasgal ( the apex telescope large area survey of the galaxy at 870@xmath0 m ) and bgps ( bolocam galactic plane survey ) @xcite , no survey at ( sub)mm wavelengths existed that trace the cold dust emission stemming from dense and young star - forming regions at adequate spatial resolution ( the cobe ( cosmic background explorer ) and wmap ( wilkinson microwave anisotropy probe ) data have too coarse resolution to isolate individual star - forming regions ) . here , we employ the 870@xmath0 m submm continuum survey atlasgal to study the general distribution of the dense dust and gas within our galactic plane . the 870@xmath0 m data are taken from the apex telescope large area survey of the galaxy ( atlasgal , @xcite ) . the @xmath4 rms of the data is @xmath350mjybeam@xmath5 and the fwhm @xmath6 . using the clumpfind source identification algorithm by @xcite with a @xmath7 threshold of 300mjybeam@xmath5 , we identified 16336 clumps within the galactic plane for longitudes @xmath1 and latitudes @xmath8 . in the context of this paper , we are not aiming for exact fluxes , column densities or masses , but we just want to evaluate source number counts within the galactic plane . therefore , the specific clump identification algorithm or the used thresholds are not of great importance . to test this , we also derived corresponding source catalogs using @xmath9 or @xmath10 thresholds . while the absolute number of sources obviously varies significantly with changed thresholds , the structural results presented below are not significantly affected by that . as an additional test , instead of deriving clumps , we just extracted the total submm fluxes above the @xmath7 threshold in the given latitude and longitude bins . again the structural distributions in longitude and latitude are very similar . since other galactic plane surveys usually also work on source counts ( e.g. , glimpse ( galactic legacy infrared midplane extraordinaire ) or msx ( midcourse space experiment ) ; @xcite ) , for the remainder of the paper we adopt the @xmath7 source catalog . the clump masses range between 100 and a few 1000m@xmath11 , and these clumps form clusters with certain star formation efficiencies . therefore , the atlasgal data largely trace gas / dust clumps capable of forming intermediate- to high - mass stars at distances between several 100 out to more than 10000pc ( @xcite ; tackenberg et al . subm . ) . tackenberg et al . ( subm . ) analyzed the atlasgal data in the longitude range between 10 and 20deg in depth via correlating them with the glimpse and mipsgal ( mips galactic plane survey ) near- to mid - infrared surveys of the galactic plane @xcite , and at long wavelength with the nh@xmath12 spectral line data from wienen et al . ( subm . ) . out of 210 starless clump candidates , tackenberg et al could extract nh@xmath12 spectral information and by that kinematic distances for 150 sources . to resolve the kinematic distance ambiguity , these targets were compared to the glimpse and mipsgal images . clumps associated with glimpse / mipsgal shadows were assigned the near distance , and the other clumps were assigned as far . this way , 115 clumps are likely on the near side of the galaxy and 35 on the far side . tackenberg et al . find that the mean distances of starless clumps in the longitude range between 10 and 20deg on the near and far side of the galaxy are 3.1 and 13.8kpc , respectively . one should keep in mind that the rotation curve of the inner galaxy is far from circular ( e.g. , @xcite ) , making the absolute determination of kinematic distances a difficult task . for comparison with somewhat more evolved evolutionary stages , namely young stellar objects ( ysos ) , we resort to the spitzer red source catalog ( with a color criterion of @xmath13-[8.0]\geq$]1 , for more details see @xcite ) . while the intrinsically red sources are contaminated by approximately 30% agb stars , @xcite extracted statistically these agb stars and provided a yso catalog with reduced agb contamination . this yso catalog contains 11649 sources again over the galactic longitude range @xmath1 . because deriving distances and masses for individual ysos is difficult , @xcite conducted a population synthesis analysis of the sample . they find that their detected sources consist mainly of intermediate- to high - mass stars ( between 3 and 25m@xmath11 ) at distances of several kpc . again dividing their sources in near and far sources with respect to the galactic center and galactic bar , they as well find mean distances in the longitude range between 10 and 20deg of 4.9 and 13.1kpc , respectively . figure [ distances ] presents a histogram of the kinematic distances in the galactic longitude range between 10 and 20deg derived for the atlasgal sample ( tackenberg et al . subm . ) and population synthesis distances for the glimpse sources @xcite , respectively . clearly , both distributions show a near and far distance peak ( see also fig . 9 in @xcite ) . for the atlasgal sources , the far peak has less sources because , at the given spatial resolution of @xmath14 , close clumps that would be spatially resolved at the near distance merge and appear as one clump at the far side of the galaxy ( tackenberg et al . subm . ) . to be more precise , tackenberg et al . ( subm . ) simulated the distance smoothing effect for the atlasgal data , and they found that an artificial sample of 328 clumps at 3kpc distance would appear as only 20 clumps when put to 15kpc distance ( see also the corresponding discussion in @xcite ) . this effect is far less severe for the much better resolved glimpse sources . additionally , glimpse sources easier saturate at the near side of the galaxy . combining these effects with the larger observed volume at the far side of the galaxy , more sources are found at the far side for that sample . although the near peaks of the atlasgal and glimpse samples are shifted with respect to each other a little bit , taking into account the inherent uncertainties of kinematic distances for the atlasgal sample and population synthesis for the yso sample ( @xmath31kpc each ) , the two source types cover comparable distance ranges . another difference between the two samples is that atlasgal barely detects any sources between 5 and 10kpc whereas the glimpse distribution shows sources there . the latter is due to the model used for the population synthesis with an axisymmetric ring around the center of the galaxy ( see fig . 2 in @xcite ) . the non - detection of atlasgal sources in that distance regime is likely attributed to the large extend of the galactic bar ( see also fig . [ sketch ] ) which is not taken into account in galactic rotation models to derive the kinematic distances ( e.g. , @xcite ) . while the atlasgal clumps are in absolute terms more massive than the glimpse ysos , considering typical star formation rates and an initial mass function for each cluster - forming region , the atlasgal clumps form intermediate- to high - mass stars in the mass range of the glimpse ysos . in combination , the two samples are well suited for comparison of the different evolutionary populations . in addition to this , since the galactic longitude range between 10 and 20 represents a fraction of the galactic plane that covers near and far spiral arms , we consider it to first order as representative to extrapolate the distance similarities for both samples also for the rest of the surveys . additional similarities and differences will be discussed in section [ lat ] . figure [ histo_lon ] presents the longitude distribution of the submm continuum and yso sources within our galactic plane ( see @xcite for a first version of such a plot based on a far smaller initial dataset ) . while the yso distribution based on the spitzer data is relatively flat , the submm continuum emission shows a series of distinctive peaks . the most prominent one is toward the galactic center where the source count increases approximately by a factor of 4 . in addition to this , there are a few more clear submm source count peaks at positive and negative longitudes . they can mainly be attributed to tangential points of spiral arms ( for a discussion of older cobe 240@xmath0 m data , see @xcite ) as well as to a few prominent star formation complexes in the nearby sagittarius arm . the most important of these are marked in figure [ histo_lon ] . for comparison , a sketch of our galaxy as it would be viewed face - on is shown in fig . [ sketch ] where several lines of sight are marked corresponding to increased number counts in fig . [ histo_lon ] . for longitudes @xmath15deg , a similar distribution was found in the bgps survey @xcite . the implications suggested by this result will be discussed further in section [ general ] . another way to represent the 870@xmath0 m source distribution is in a 2-dimensional binning in galactic longitude and latitude ( fig . [ lon_lat ] top panel ) . in addition to an increase in source counts toward specific galactic longitudes , we also identify a tight confinement to the galactic mid - plane with only a narrow spread north and south of the mid - plane . to derive the approximate latitude for which the submm emission peaks in each longitude bin , we fitted gaussians to their latitude distribution in each longitude bin . the gaussian fit peak positions are marked in fig . [ lon_lat ] ( top panel ) . these fits indicate that the dominant dust and gas distribution is slightly shifted to negative galactic latitudes with a mean offset over the whole plane of @xmath16deg ( the mean values are derived from gaussian fits to 10deg longitude bins , see below ) . + we can produce the same plot for the yso distribution derived from the spitzer data ( fig . [ lon_lat ] bottom panel ) . similarly , the mean value of the peaks of the yso distribution is also shifted below the galactic mid - plane , again at @xmath17deg ( the mean values are also derived from gaussian fits to 10deg longitude bins ) . already a visual inspection of the two distributions indicates that the ysos appear to cover a broader range in galactic latitude than the dust and gas clumps traced by the submm continuum emission . fitting gaussians to the latitude distributions in each longitude bin for the submm clumps as well as the ysos allows us to better quantify this effect . since the latitude distributions are not as smooth on the scales of individual degrees in galactic longitude , we average over 10 deg in longitude for smoothing purposes . figure [ gauss ] presents gaussian example fits at different galactic longitudes outlining the applicability of the gaussian assumption to these distributions . the corresponding gaussian full width half maximum ( fwhm ) for the two distributions are shown in figure [ width ] . one clearly sees that the yso distributions is broader over the whole galactic plane than the dense gas and dust distribution . below galactic longitudes of -30deg , the atlasgal distribution shows a tendency of increased fwhm . however , this effect is confined to only three bins , one with a particularly large error - bar . therefore , in the context of this paper we refrain from further interpretation . the mean values of the fwhm for the dust continuum and yso distributions are @xmath30.67@xmath18 and @xmath31.09@xmath18deg , respectively . this corresponds to characteristic scale heights @xmath19 ( distance where distribution has dropped to 1/e , @xmath200.6@xmath21fwhm ) of @xmath30.4 and @xmath30.7deg for the submm continuum and yso distributions , respectively . @xcite fitted their source distribution with a mean physical scale height of 80pc . since we have no explicit distances for individual atlasgal sources over the whole range of the galactic plane , deriving directly a physical scale height from our data alone is hardly feasible . however , because the distance and mass regimes of the atlasgal sources and the ysos are similar ( see section [ obs ] ) , we can use the linear scale height of the ysos to derive a first - order estimate of the linear scale height for the atlasgal clumps ( see section [ lat ] for additional discussion ) . with the measured fwhms of the ysos and the atlasgal sources , and using the modeled 80pc mean physical scale height of the ysos , we estimate , proportionally , an approximate physical scale height of the submm dust continuum emission of 46pc . since the distribution of near and far sources favors near sources for atlasgal and far sources for glimpse ( see fig . [ distances ] ) , the 46pc should be considered as an upper limit . + + another estimate of the atlasgal scale height can be derived from the sample discussed in section [ obs ] for the galactic longitude range between 10 and 20deg ( tackenberg et al . subm ) . using only the near kinematic distance sample ( the far distance sample has too low number statistics ) , we find a fwhm and scale height of 23 and 14pc , respectively . this is considerably lower than the estimate of 46pc derived above . the most important reason for this difference is based on the way near and far distance sources are separated in the study by tackenberg et al . ( subm . ) : while the kinematic distances always give near and far solutions at this and the other side of the galaxy , the near solution was favored for infrared dark clouds if absorption shadows were detected in the spitzer mid - infrared data ( e.g. , @xcite ) . however , at high galactic latitudes the infrared background is significantly reduced and by default it is more difficult to identify infrared dark clouds . therefore , from a statistical point of view , at high latitude , clumps are potentially shifted to the far distance . this implies that the derived scale height of 14pc from this sample has to be considered as a lower limit . to first order , it appears surprising that the cold dust clump distribution and the yso distribution do not resemble each other more closely . while the submm continuum emission is optically thin and should therefore trace almost all cold dust along each line of sight , the yso distribution is more strongly affected by extinction . it may well be that a considerable fraction of ysos are not identified because of too high extinction . regarding the different properties of the atlasgal clumps and ysos toward the galactic center ( fig . [ histo_lon ] ) , a similar emission increase toward the galactic center was recently also found in the dense gas emission of nh@xmath12 ( the hops survey , e.g. , @xcite , purcell et al . , longmore et al . in prep . ) . the lack of a prominent peak toward the galactic center in the glimpse yso data may be partly an observational bias but also partly a real physical effect . observationally , the extinction toward the galactic center increases which increases the glimpse detection threshold . however , in contrast to that , also surveys of hii regions , tracing more evolved high - mass star - forming regions , as well as h@xmath22o and ch@xmath12oh maser surveys exhibit no strong emission peaks toward the galactic center ( e.g. , @xcite ) . does that imply a relatively low gas - to - star conversion in that specific part of our galaxy ? for a more detailed discussion , see longmore et al . in prep .. @xcite presented a similar plot to our fig . [ histo_lon ] for the distribution of irdcs in the galactic plane . while the general features for the irdcs appear similar to the dust continuum distribution , the spiral arm and star - forming regions are less pronounced . @xcite performed a similar study of the hii region distribution of the galactic plane accessible to the northern hemisphere , whose results are noticeably different to what we find . the hii region distribution does not exhibit a peak toward the galactic center region , but a clear peak is found at approximately + 30deg , corresponding to our peak for the scutum arm . previous hii region surveys show a similar hii number increase at negative longitudes around -30deg @xcite . recently , @xcite report on the ch@xmath12oh maser distribution in the galactic plane between longitude @xmath23deg . among other source count peaks , in particular they report an increased detection rates at longitudes around @xmath24 and @xmath25deg , very similar to what we find . while the peaks at @xmath24 and @xmath26 deg are likely associated with the end of the long galactic bar and the beginning of the scutum - centaurus spiral arm , the @xmath25deg peak should be associated with a tangent point of the 3kpc arm ( fig . [ sketch ] ) . in summary , the submm continuum emission is an excellent tracer of the galactic dense gas structure , even in our milky way where our location within the plane complicates the picture so severely . the finding that the average peak of the gaussian latitude distribution is below the galactic plane has already been inferred by other groups , e.g. , for ( sub)mm continuum emission @xcite , in the infrared @xcite , for co emission @xcite , clusters @xcite , hii regions @xcite or infrared bubbles ( kendrew priv . , simpson et al . in prep . ) . even the galactic center itself is located at 0.05deg below the galactic plane @xcite . while a common interpretation of that effect is based on a poor definition of the galactic plane where neither the sun nor the galactic center itself are located directly in the plane at 0deg latitude ( e.g. , @xcite ) , @xcite recently suggested that this effect may also simply be caused by individual star formation complexes and not so much reflect a global galactic property . however , they state that the offset is mainly a feature of the molecular interstellar medium , whereas different studies of the glimpse survey indicate that the stellar component shows the same effect ( e.g. , @xcite , this study ) . although we can not conclusively differentiate these scenarios , the data here are indicative of a real global offset of the galactic midplane from its conventional position where the axis between the sun and the galactic center are located at @xmath27deg . are the different galactic latitude distributions of the submm clumps and the ysos a real physical effect or could they be caused by observational biases ? as outlined in section [ obs ] , the mass distributions and the distances on the near and far side of the galaxy of the two samples are similar . one may now ask whether the number of near sources were larger for the ysos than for the submm clumps . however , there are several effects that counteract this : at the relatively coarse spatial resolution of atlasgal ( @xmath14 ) , clumps that would be separate entities on the near side of the galaxy merge into single objects on the far side , and the total number of sources on the far side is lower than that on the near side ( tackenberg et al . in contrast to that , at the higher spatial resolution of spitzer ( @xmath28 ) , most sources remain point sources independent of the distance , and therefore less suffer from the `` merging - problem '' . furthermore , the observed volume at the far side of our galaxy is larger than that on the near side , and glimpse sources on the near side more easily saturate . these combined effects even cause a yso number increase on the far side compared to the near side of our galaxy ( fig . [ distances ] , @xcite ) . therefore , the mass and distance distributions of the submm clumps and ysos are unlikely to be the cause for the difference in the latitude distribution . furthermore , @xcite statistically excluded the agb star population from their catalogue , which implies that contamination by post - main - sequence sources is not responsible for the difference in latitude distribution as well . similar to the effect seen for the longitude distribution discussed in the previous section ( [ lon ] ) , extinction may cause part of the broader latitude distribution because toward the highest extinction regions , that are traced by the submm continuum emission , infrared emission is hardly detectable and hence the most deeply embedded yso population is likely to be missed by the spitzer data or the selection criteria in @xcite . herschel longer - wavelength data may identify such embedded population better ( e.g. , @xcite ) . @xcite recently reported a galactic average scale height for h@xmath22o masers of approximately 0.4deg , earlier finding a scale height for the ch@xmath12oh class ii masers in a similar range @xcite . assuming a comparable distance distribution for the maser as well as the submm continuum sources , the mean physical scale - height of the masers should also be @xmath2946pc . for ultracompact hii regions , @xcite found a scale height of 0.6deg , intermediate between our values for the submm clumps and the ysos . later , @xcite reported a smaller mean physical scale - height for ultracompact hii regions of @xmath330pc ( about 40% of the @xcite value ) , claiming that the @xcite sample is biased by its large fraction of b - stars . similar mean physical scale heights for high - mass star - forming regions of @xmath344pc and @xmath329pc were reported by @xcite and @xcite . hence masers as typical tracers of star - forming regions ( a fraction of the h@xmath22o masers may also stem from agb stars ) , dust continuum emission as a tracer of the dense gas , and young high - mass stars exhibit similar scale height distributions in the milky way . in comparison , @xcite find an approximate scale height of 70pc for co ( rescaled to a galactocentric solar distance of 8.5kpc ) , and @xcite derive a values of @xmath3120pc for the thick co disk ( which is an average of their 3 cited values ) . the reported cold hi scale height is around 150pc @xcite . therefore , while tracers of the youngest evolutionary stages of star formation ( submm continuum emission and masers ) are all found closest to the galactic plane , more evolved evolutionary stages like ysos as well as the less dense atomic and molecular gas appear to be located on average slightly further from the plane . using the estimated mean physical scale heights for the dust continuum and yso distributions of 46 and 80pc , respectively , we can calculate approximate velocities required to move the 30pc difference in the given yso lifetimes of 12myrs . this estimate results in required yso velocities between 15 and 30kms@xmath5 . while velocities of that order are found ( e.g. , pv cephei , @xcite ) , they are apparently not the rule . therefore , we propose that the most likely explanation for the spread in scale height for different populations appears to be a combination of extinction effects and dissolving young clusters from their natal birth sites . we present a study of the galactic distribution of submm dust continuum emission from the northern and southern hemisphere . the submm continuum emission , which traces the dense gas emission from star - forming regions , excellently traces the structure within our galaxy . spiral arms and prominent structures of star formation are easily distinguished by significantly increased source counts toward these galactic longitudes . the galactic latitude distribution is skewed slightly below the galactic plane , and the mean physical scale height is estimated to 46pc . this scale height corresponds well to other star - formation tracers like ch@xmath12oh and h@xmath22o maser emission , and it is more confined to the galactic plane than most of the other populations in our galaxy . we compare the submm continuum emission with several other tracers of galactic structure , in particular with the yso population identified by spitzer observations . the yso population has a significantly larger scale height , and we propose that this may be attributed to a combined effect of extinction and dissolving of protostellar clusters after their birth . we thank the referee for a careful review improving the paper significantly . likes to thanks sarah kendrew for discussion about galactic distributions . lb acknowledges support from conicyt through projects fondap no . 15010003 and basal pfb-06 .	the atlasgal 870@xmath0 m continuum survey conducted with the apex telescope is the first survey covering the whole inner galactic plane ( @xmath1 & @xmath2 ) in submm continuum emission tracing the cold dust of dense and young star - forming regions . here , we present the overall distribution of sources within our galactic disk . the submm continuum emission is confined to a narrow range around the galactic plane , but shifted on average by @xmath30.07deg below the plane . source number counts show strong enhancements toward the galactic center , the spiral arms and toward prominent star - forming regions . comparing the distribution of atlasgal dust continuum emission to that of young intermediate- to high - mass young stellar objects ( ysos ) derived from spitzer data , we find similarities as well as differences . in particular , the distribution of submm dust continuum emission is significantly more confined to the plane than the yso distribution ( fwhm of 0.7 and 1.1deg , corresponding to mean physical scale heights of approximately 46 and 80pc , respectively ) . while this difference may partly be caused by the large extinction from the dense submm cores , gradual dispersal of stellar distributions after their birth could also contribute to this effect . compared to other tracers of galactic structure , the atlasgal data are strongly confined to a narrow latitude strip around the galactic plane .
queueing processes have been studied extensively , especially due to their practical relevance @xcite . however , usually the spatial structure of the queues is neglected and the particles in the queues do not interact with each other . on the other hand , the totally asymmetric simple exclusion process ( tasep ) which has a spatial structure and excluded - volume effect ( hard - core repulsion ) is one of the best - studied interacting particle systems @xcite . nowadays the tasep is a basic model for pedestrian and traffic flows @xcite . recently a queueing process with the excluded - volume effect , the exclusive queueuing process ( eqp ) , was introduced in @xcite and @xcite independently , where the model was formulated as continuous - time and discrete - time markov processes , respectively . this model can be rephrased as the tasep on a semi - infinite chain with a new boundary condition , see fig . [ fig : eq ] . the left end is interpreted as the end of the queue where new customers arrive . therefore particles can enter at the left site next to the leftmost occupied site . the ( fixed ) rightmost site corresponds to the server . here particles can leave the system after getting service . in the bulk a particle can hop to its right nearest neighbor site , if the target site is empty . after the study @xcite , where the bulk hopping rule is deterministic , the model with discrete time and probabilistic bulk hopping was analyzed @xcite . earlier works @xcite focussed on the exact probability distribution , physical quantities in the stationary state , and the conditions under which the eqp converges to the stationary state . in this paper , we study the dynamical properties by considering the number of particles and the system length which is defined as the position of the leftmost particle . we use the same formalism as in @xcite , i.e. discrete time and parallel - update scheme . for generic hopping probability @xmath4 , we will introduce a domain wall prediction , checking it by monte carlo simulations . in the deterministic hopping case @xmath5 , a rigorous analysis is available , by using the generating function technique @xcite . this paper is organized as follows . in sec . [ model ] , we define the model as a discrete - time markov process , and review its stationary state based on @xcite , which can be generalized to the inhomogeneous injection case , see also app . [ inhomo ] . in sec . [ domain - wall ] , we introduce a phenomenological argument on how the number of particles @xmath0 and the system length @xmath3 converge or diverge , showing simulation results . in sec . [ p=1 ] , we derive the asymptotic behaviors of @xmath0 and @xmath3 rigorously for @xmath5 , imposing the initial condition that there is no particle on the chain . section [ conclusion ] is devoted to the conclusion of this paper . the eqp is defined on a semi - infinite chain where sites are labeled by natural numbers from right to left ( fig . [ fig : eq ] ) . particles can enter the chain with probability @xmath6 only at the left site next to the leftmost occupied site . a particle hops to its right nearest neighbor site with probability @xmath4 , if it is empty , and exits at the right end of the chain with probability @xmath7 . if there is no particle on the chain , a particle enters at site 1 with probability @xmath6 . these transitions occur simultaneously within one time step , i.e. we apply the fully - parallel - update scheme . we formulate the eqp as a discrete - time markov process on the state space @xmath8 where 0 and 1 correspond to unoccupied and occupied sites , respectively . in particular , @xmath9 denotes the state in which there is no particle on the chain . to simplify the notation we do not write the infinite number of 0 s located left to the leftmost 1 . let us review the matrix product stationary state @xcite of the model , which is a simple extension of that for systems with a fixed system length @xcite . when @xmath10 the stationary state can be expressed as @xmath11 @xmath12 and @xmath13 are matrices , @xmath14 is a row vector and @xmath15 is a column vector satisfying the algebraic relations @xmath16 these relations are closely related to those for the stationary state of the parallel - update tasep with ordinary open boundary condition @xcite . the normalization constant is expressed as @xmath17 with @xmath18 . the average number of particles @xmath19 and the average system length ( the position of the leftmost particle ) @xmath20 are calculated as @xmath21 in the stationary state . note that @xmath22 and @xmath23 diverge on the critical line @xmath24 , @xmath25 , where the stationary state exists . a generalization of the model , where the entry probability depends on the system length , also has a matrix product stationary state , see app . [ inhomo ] . in this section , we discuss the time evolution of the average number of particles @xmath0 and the average system length @xmath3 corresponding to the position of the leftmost particle . in the ordinary open boundary case , where the length of the system is fixed , a domain wall moves rightward or leftward , or exhibits a random walk depending on the boundary parameters @xcite . in the same way , we will discuss how the system length @xmath3 moves . we also observe how the average number of particles @xmath0 changes as well . the continuity equation @xmath26 holds , where @xmath27 and @xmath28 are the flows of particles entering and leaving the system , respectively . the inflow @xmath27 is always @xmath6 , which is due to the fact that the site where particles enter is by definition never blocked . in other words , our model is not a call - loss system . under the assumption that the outflow @xmath29 is independent of @xmath30 , we have @xmath31 . in fact our simulations show that both @xmath32 and @xmath3 decrease or increase linearly in time @xmath30 according to @xmath33 or @xmath34 , respectively . when @xmath33 , the system converges to the stationary state , . we impose the initial condition that particles are distributed uniformly with density @xmath35 as in fig . [ fig : sim - conv ] , @xmath32 and @xmath36 decrease linearly in time as @xmath37 with @xmath38 since @xmath39 , we have @xmath40 which means that the domain wall moves rightward . and @xmath41 , and the initial conditions as @xmath42 and @xmath43 ( with @xmath44 defined by ) , respectively . the green and red lines are data for @xmath0 and @xmath3 obtained from monte carlo simulations , where 5000 samples are averaged . the black and blue lines correspond to the predictions of the domain wall theory . note that the asymptotic values are small but non - zero ( see eqs . and ): @xmath45 and @xmath46 , respectively . , title="fig : " ] + and @xmath41 , and the initial conditions as @xmath42 and @xmath43 ( with @xmath44 defined by ) , respectively . the green and red lines are data for @xmath0 and @xmath3 obtained from monte carlo simulations , where 5000 samples are averaged . the black and blue lines correspond to the predictions of the domain wall theory . note that the asymptotic values are small but non - zero ( see eqs . and ): @xmath45 and @xmath46 , respectively . , title="fig : " ] + according to the forms for @xmath47 and @xmath44 , we call the phases @xmath48 maximal - current - convergent ( mc - c ) and high - density - convergent ( hd - c ) phases , respectively , see fig . [ fig : subphase ] . + it should be noted that the outflow is given by eqn . only while @xmath49 . as @xmath50 , the outflow approaches @xmath6 , assuring that @xmath0 approaches the stationary value . figure [ fig : sim - conv - rho ] shows density profiles in the hd - c and mc - c phases . we can observe that the bulk density keeps its initial value . of the @xmath51th site at time @xmath30 ) in the hd - c ( top ) and mc - c ( bottom ) phases . the parameters @xmath52 and the initial conditions @xmath53 are set to the same values as in fig . [ fig : sim - conv ] . the colored snapshots are obtained by averaging 5000 samples of monte carlo simulations . the black lines represent the predicted densities @xmath44 according to eqn . . , title="fig : " ] + of the @xmath51th site at time @xmath30 ) in the hd - c ( top ) and mc - c ( bottom ) phases . the parameters @xmath52 and the initial conditions @xmath53 are set to the same values as in fig . [ fig : sim - conv ] . the colored snapshots are obtained by averaging 5000 samples of monte carlo simulations . the black lines represent the predicted densities @xmath44 according to eqn . . , title="fig : " ] + and @xmath54 , respectively , and the initial conditions as @xmath55 ( with @xmath44 defined by ) . the green and red lines are data for @xmath0 and @xmath3 obtained from monte carlo simulations , where 1000 samples are averaged . the black and blue lines correspond to the predictions of the domain wall theory . , title="fig : " ] + and @xmath54 , respectively , and the initial conditions as @xmath55 ( with @xmath44 defined by ) . the green and red lines are data for @xmath0 and @xmath3 obtained from monte carlo simulations , where 1000 samples are averaged . the black and blue lines correspond to the predictions of the domain wall theory . , title="fig : " ] + when @xmath34 , it is natural to expect that the domain wall moves leftward , and the time evolutions of @xmath0 and @xmath3 are expressed by eqs . and , respectively , with the density and the outflow . the simulations imply that this is true for @xmath32 with @xmath56 but fails for @xmath36 , see fig . [ fig : sim - div ] . this failure is not unexpected since the predicted velocity @xmath57 can be greater than 1 whereas the length @xmath36 can not be larger than @xmath58 by the definition of the model . however , the simulation results ( see fig . [ fig : sim - div ] ) indicate that @xmath59 so that the prediction is qualitatively correct . the velocity @xmath60 satisfies @xmath61 and we have exactly @xmath62 when @xmath63 . moreover , when @xmath5 , we will show in the next section that @xmath64 equations and can be regarded as the asymptotic behaviors @xmath65 in the same way as in the convergent phase , we call the subphases @xmath66 maximal - current - divergent ( mc - d ) and high - density - divergent ( hd - d ) phases , respectively . note that the densities in the mc - d and hd - d phases are higher than ( or equal to ) @xmath67 and @xmath68 , respectively . it is difficult to predict how @xmath0 or @xmath3 behaves just on the critical line @xmath69 . for @xmath5 , however , we will find in the next section diffusive behavior on the critical line as @xmath70 with constants @xmath71 and @xmath72 . .,title="fig : " ] + in this section we investigate the asymptotic behaviors of @xmath73 and @xmath3 rigorously for @xmath5 . thanks to the deterministic particle hopping we can obtain the generating functions of @xmath0 and @xmath3 . for simplicity , we impose the initial condition @xmath9 ( there is no particle in the system at time @xmath74 ) , and reset the state space as @xmath75 note that , for @xmath5 , the sequence 00 never appears if the system starts from the initial condition @xmath9 . in this case , the mc - d and mc - c phases vanish from the phase diagram , see fig . [ fig : pd184 ] . we first consider the number of particles , borrowing the classification from @xcite as @xmath76 , \end{split } \\ \begin{split } p_t^b(n ) = & { \rm prob}\left[\begin{array}{l } \text{\ # of particles is $ n$ at time $ t$ } \\ \wedge \text { site 1 is empty at time $ t$ } \end{array } \right ] \end{split}\end{aligned}\ ] ] for @xmath77 with @xmath78 . these probabilities are governed by the following master equation , which was found in @xcite : @xmath79 this simplification is due to the deterministic hopping @xmath5 . we choose the initial condition such that @xmath80 we will check that the average number of particles @xmath73 converges to the stationary value when @xmath81 , and show that @xmath0 behaves as eqn . when @xmath82 . we will also show that @xmath73 exhibits diffusive behavior on the critical line @xmath83 . we define the generating functions of @xmath84 and @xmath85 as @xmath86 for @xmath87 . noting the initial condition , we find @xmath88 from eqn . , we have @xmath89 inserting this and eqn . into eqs . and , we get a recurrence formula for @xmath90 as @xmath91 the recurrence formula has the following solution : @xmath92 where @xmath93 with @xmath94 are solutions to @xmath95 with @xmath96 and @xmath97 as defined in eqn . . restricting the `` initial condition '' @xmath98 and @xmath99 such that @xmath100 we have @xmath101 ( note that @xmath102 . ) from this constraint and the relation @xmath103 is determined , and we find @xmath104 then we obtain the generating function @xmath105 of the probability that the number of particles is @xmath106 as @xmath107 we also introduce the generating function @xmath108 of the generating function @xmath105 as @xmath109 and @xmath110 of the average number of particles as @xmath111 the asymptotic behavior of @xmath0 is determined by the degree of the singularity @xmath112 of @xmath110 @xcite . when @xmath81 , we find @xmath113 and @xmath114 converges as @xmath115 of course this limit value agrees with the stationary value with @xmath5 . when @xmath82 , we find @xmath116 and @xmath114 behaves as @xmath117 when @xmath83 , we find @xmath118 and @xmath114 behaves as @xmath119 now we turn to the behavior of the length of the system ( the position of the leftmost particle ) . let @xmath120 be the probability that the system length is @xmath121 at time @xmath30 . the probability @xmath120 is governed by @xmath122 for @xmath123 . the first equation means that , if there is no particle at time @xmath124 , there is no particle at time @xmath30 and no particle enters ( with probability @xmath125 ) , or there is only one particle on the rightmost site at time @xmath30 which leaves the system and no particle enters ( with probability @xmath126 ) . the second equation is derived in app . [ derive - ql ] . in the same way as for the number of particles , we define the generating function @xmath127 . noting the initial condition @xmath128 and @xmath129 , we find @xmath130 the solution to the recurrence formula is @xmath131 where @xmath132 are the solutions to @xmath133 with @xmath134 and @xmath135 as defined in eqn . . due to the condition @xmath136 , the `` initial condition '' must be restricted as @xmath137 ( note that @xmath138 . ) thus we find @xmath139 the generating function of the generating function is calculated as @xmath140 and that of the average system length as @xmath141 the asymptotic behavior of @xmath3 is determined by the degree of the singularity @xmath112 of @xmath142 . when @xmath81 , we find @xmath143 and @xmath144 converges as @xmath145 again this limit value agrees with the stationary value with @xmath5 . when @xmath82 , we find @xmath146 and @xmath144 behaves as @xmath147 when @xmath83 , we find @xmath148 and @xmath144 behaves as @xmath149 we have investigated the dynamical properties of the eqp , a queueing process with excluded - volume effect . the model can be interpreted as a tasep with varying length . using generating function techniques and a phenomenological domain wall theory we have derived analytical predictions for the time - dependence of the number of particles @xmath0 and the average system length @xmath150 . we found that the two phases observed previously can be divided in subphases . the convergent phase , where the system length remains finite , consists of high - density and maximal current subphases . the same is true for the diverging phase , where the system length becomes infinite in the long time limit . by comparing with monte carlo simulations it was found that the predications of the domain wall theory for the dynamical behavior are at least qualitatively correct , i.e. @xmath3 and @xmath0 converge or diverge linearly in time . moreover they are in good agreement even quantitively in the convergent phase . in the divergent phase , the predicted velocity for @xmath73 appears to be correct whereas deviations from the domain wall theory can be observed for @xmath3 . for @xmath5 , we derived exact analytical results for the behaviors of @xmath0 and @xmath3 by using the generating function method . we showed the linearity of their dynamics in the divergent phase as predicted by the domain wall theory . we found diffusive behavior on the critical line as well . the simple approach presented here does not provide a good expression for the velocity @xmath60 of the system length ( @xmath151 ) in the divergent phase . here further analysis of the detailed density profiles in the divergent phase may be helpful . a first step would be the numerical determination of @xmath60 to get a better understanding of its dependence on the parameters @xmath152 and @xmath4 . another way to settle the problem is extending the exact result to @xmath153 case , which may be difficult but very worthwhile . here we consider the stationary state for a generalized model where the entry probability depends on the system length . a new particle enters the system with probability @xmath154 if the leftmost occupied site is @xmath121 , or @xmath155 if there is no particle on the chain . the stationary state of this generalized model can be written in the following matrix product form with the same matrices and vectors ( @xmath156 and @xmath157 ) : @xmath158 this can be proved in the same way as in the homogeneous case @xmath159 , see @xcite . to derive eqn . , we show @xmath160 , \end{split } \\ \begin{split } \label{1-betaq } & ( 1-\beta ) q_t(l ) \\ & = { \rm prob}\left[\begin{array}{l } \text{the system length is $ l$ at time $ t$ } \\ \wedge \text { ( $ l-1$)th site is occupied at time $ t$ } \end{array } \right ] , \end{split}\end{aligned}\ ] ] for @xmath161 in the deterministic hopping case @xmath5 with the initial condition @xmath9 , the sequence 00 ( except the infinite number of 0 s left to the leftmost particle ) never appears , and holes necessarily `` hop '' leftward . thus the first site must be occupied by a particle at time @xmath163 if the system length at time @xmath30 is @xmath121 . the local state ( empty or occupied ) of the @xmath164th site at time @xmath30 depends only on whether the first particle exits or not at time @xmath163 . let the number of particles be @xmath165 at time @xmath163 , and label the particles by natural numbers as in fig . [ transition - example ] . we can show by induction that the @xmath166th particle is on the @xmath166th site at time @xmath167 . thus we find that @xmath168 new particles should enter the system during @xmath169 $ ] . we obtain ( the summation of ) the transition probability from a configuration @xmath170 at time @xmath163 to the states in which the system length is @xmath121 and the @xmath164th site is empty or occupied at time @xmath30 is given by @xmath171 respectively , where @xmath172 , @xmath173 and @xmath174 denote the transition probability from @xmath175 to @xmath176 , the length of the state @xmath175 and the local state of site @xmath51 . the binomial @xmath177 gives the number of possibilities for when the new particles enter the system . these equations lead to eqs . and . * the length is @xmath178 and a new particle enters ( with probability @xmath6 ) , * the length is @xmath121 , the ( @xmath178)th site is occupied , and no particle enters ( with probability @xmath125 ) , * the length is @xmath179 , the @xmath121th site is empty , and no particle enters ( with probability @xmath125 ) . the authors thank alexandru aldea , philip greulich , joachim krug and gunter m. schtz for useful discussion . this work is supported by grant - in - aid for young scientists ( ( b ) 22740106 ) and global coe program `` education and research hub for mathematics - for - industry . ''	in a recent study @xcite the stationary state of a parallel - update tasep with varying system length , which can be regarded as a queueing process with excluded - volume effect ( _ exclusive queueing process _ , eqp ) , was obtained . we analyze the dynamical properties of the number of particles @xmath0 and the position of the last particle ( the system length ) @xmath1 , using an analytical method ( generating function technique ) as well as a phenomenological description based on domain wall dynamics and monte carlo simulations . the system exhibits two phases corresponding to linear convergence or divergence of @xmath2 and @xmath3 . these phases can both further be subdivided into high - density and maximal - current subphases . the predictions of the domain wall theory are found to be in very good agreement quantitively with results from monte carlo simulations in the convergent phase . on the other hand , in the divergent phase , only the prediction for @xmath0 agrees with simulations .
the _ integer quantum hall effect _ ( iqhe ) reveals a variety of surprising and attractive physical features , and has been the subject of several investigations ( see @xcite and references therein ) . in fact , a complete spectral analysis of the schrdinger operator for a single particle moving in a plane in a periodic potential and subject to an uniform orthogonal magnetic field of strength @xmath0 ( _ magnetic bloch electron _ ) is extremely difficult . thus the need for simpler effective models which hopefully capture ( some of ) the main physical features in suitable physical regimes . in the limit of a strong magnetic field , @xmath1 , the iqhe is well described by an effective harper operator ( cf . @xcite ) . for this model the quantization ( in units of @xmath2 ) of the hall conductance has a geometric meaning being related to chern numbers of suitable naturally bundles associated to spectral projections of the operator . a family of diophantine equations , the tknn - equations of @xcite , provides a recipe for computing such integers . the aim of the present paper is to derive a generalized version of the tknn - equations yielding the hall conductances for more general dirac - like harper operators . interest in such generalizations comes also from these dirac - like operators appearing naturally in important physical models , notably models for the graphene . with @xmath3 , the effective harper operator is @xmath4 acting on the hilbert space @xmath5 . that this operator is the simplest representative of a large family of generalized harper operators , sharing similar mathematical properties , is our starting point . on the hilbert space @xmath6 consider the unitary operators @xmath7 with @xmath8 . they are readily seen to obey the relation @xmath9 yielding for the harper operator the expression @xmath10 . for any positive integer @xmath11 on the vector space @xmath12 consider two unitary @xmath13 matrices @xmath14 and @xmath15 defined as follows . let @xmath16 be the canonical basis of @xmath12 , then @xmath14 is a diagonal matrix and @xmath15 is a shift matrix acting as @xmath17_q } , \ ] ] where @xmath18_q$ ] stays for modulo @xmath19 . they obey @xmath20 then , on the hilbert space @xmath21 one defines a pair of unitary operators : @xmath22 with @xmath23 and @xmath24 and @xmath25 given by ( [ eq_004 ] ) . the integer @xmath26 is chosen coprime with respect to @xmath19 . as for the case before ( when @xmath27 ) , the operators ( [ eq_006 ] ) also obey the relation ( [ nct ] ) : @xmath28 following the definition ( [ eq_001 ] ) we can introduce the _ generalized @xmath29-harper operator _ @xmath30 more generally one considers the collection @xmath31 of bounded operators on the hilbert space @xmath32 generated by the unitaries @xmath33 and @xmath34 . technically @xmath31 is a @xmath35-algebra , i.e. an involutive algebra closed with respect to the operator norm , and it is named the _ ( ir)rational rotation algebra _ or the _ noncommutative torus algebra _ @xcite . by writing the operator in ( [ eq_001 ] ) as @xmath36 with @xmath37 in particular , the generalized @xmath38-harper operator @xmath39 is just @xmath40 acting on @xmath41 . this operator provides an interesting effective model for the iqhe on graphene @xcite . moreover , dirac - like operators like @xmath42 can be used to describe effective models for electrons interacting with the periodic structure of a crystal through a periodic ( internal ) magnetic field and subjected to the action of an external strong magnetic field @xcite . for rational deformation parameter , @xmath43 ( @xmath44 and @xmath45 taken to be coprime here and after ) , any family of operators @xmath31 can be decomposed in a continuous way according to a generalized version of the bloch theorem . more explicitly we have the following . * proposition a. * _ let @xmath43 . for any ( admissible ) pair @xmath29 the bounded operator algebra @xmath31 on @xmath32 admits a _ bundle representation _ @xmath46 over the ordinary two - torus @xmath47 . that is to say , there is a hermitian vector bundle @xmath48 together with a unitary transform @xmath49 such that @xmath50 the vector bundle @xmath51 has rank @xmath45 and ( first ) chern number @xmath52 . _ here @xmath53 denotes the hilbert space of square integrable sections of the vector bundle @xmath51 and @xmath54 denotes the collection of continuous sections of the endomorphism bundle @xmath55 , i.e. the vector bundle with fibers @xmath56 associated with the vector bundle @xmath51 . the unitary map @xmath57 implementing the bundle representation of @xmath31 is called _ ( generalized ) bloch - floquet transform _ @xcite . for the details of the proof of prop . a , that we briefly sketch , we refer to @xcite ( see also @xcite ) . denote with @xmath58 , @xmath59 the unique solution of @xmath60 ( due to @xmath19 and @xmath61 being coprime ) and be @xmath62 . then , a simple check shows that the unitary operators @xmath63 commute , @xmath64=0 $ ] , while commuting with any element in @xmath31 . they generate a ( indeed maximally ) commutative sub - algebra of the commutant of @xmath31 , and in particular , of symmetries for the operator ( [ eq_001_bis ] ) . were @xmath45 a multiple of @xmath19 this commutative sub - algebra would reduce to a direct sum of @xmath19 copies of a commutative algebra on @xmath65 . thus to avoid this degeneracy , we take @xmath19 and @xmath45 to be coprime as well . this entails there exist two integers @xmath66 and @xmath67 such that @xmath68 , a fact we shall exploit momentarily . moreover , the commutative algebra generated by @xmath69 and @xmath70 is isomorphic to the algebra of continuos functions over the ordinary 2-torus @xmath47 . a generalized simultaneous eigenvectors of @xmath69 and @xmath70 is a @xmath71 ( @xmath72 is the space of tempered distributions ) such that @xmath73 for any @xmath74 ^ 2\simeq{\mathbb t}^2 $ ] , the generalized eigenvectors make up a @xmath45-dimensional space , a basis of which being given by a fundamental family of distribution @xmath75 , for indices @xmath76 , with elements @xmath77 , @xmath78 , defined by @xmath79 .\ ] ] here the permutation @xmath80 of the set @xmath81 is defined by @xmath82_q$ ] and , as usual , the dirac delta function @xmath83 acts on functions @xmath84 as the evaluation at the point @xmath85 , i.e. @xmath86 . we let @xmath87 denote the @xmath45-dimensional vector space spanned by the distributions @xmath88 . the total space of the vector bundle @xmath51 is just the disjoint union of the spaces @xmath89 glued together with transitionf functions coming from pseudo - periodic conditions satisfied by the @xmath90 s . indeed , from ( [ eq_sec_har_01bis ] ) one deduces that @xmath91 while @xmath92 for @xmath93 and @xmath94 . also , there is an identification @xmath95 which is very reminiscent of the usual direct integral decomposition of the bloch theory . we stress that prop . a does not only states that any @xmath96 can be decomposed as a direct integral operator @xmath97 with @xmath98 an @xmath99 matrix acting on @xmath89 , but also that such a decomposition is continuous with respect to the topology of the vector bundle @xmath51 , thus amounting to a bundle representation . for @xmath100 , we denote @xmath101 . for the generators , when acting on the basis @xmath102 one finds @xmath103 here @xmath104 is the diagonal matrix @xmath105 ; @xmath106 is the twisted shift matrix sending @xmath107 to @xmath108 for @xmath109 while @xmath110 to @xmath111 . the matrices in ( [ matu - v ] ) commute up to @xmath112 being @xmath113 as before , @xmath114 , thus providing a representation of @xmath31 . moreover , their pseudo - periodic conditions @xmath115 match those of the basis @xmath102 thus making @xmath46 a representation of @xmath31 as bundle endomorphisms , as expressed in . the bundle @xmath51 comes equipped with the _ berry connection _ @xmath116 its curvature @xmath117 is constant , @xmath118 ( up to an exact form ) and when integrated it results in the first chern number of the bundle being @xmath119 for a rational @xmath43 , the spectrum of @xmath120 in ( [ eq_001_bis ] ) has @xmath121 energy bands if @xmath45 is odd or @xmath45 energy bands if @xmath45 is even @xcite . these include the _ inf - gap _ ( from @xmath122 to the minimum of the spectrum ) and the _ sup - gap _ ( from the maximum of the spectrum to @xmath123 ) . to each gap @xmath124 one associates a spectral projection @xmath125 with the convention that @xmath126 for the inf - gap @xmath127 and @xmath128 for the sup - gap @xmath129 with @xmath130 or @xmath131 according to whether @xmath45 is odd or even . as usual , the projection @xmath125 is defined via the riesz formula for the operator @xmath120 , @xmath132 the closed rectifiable path @xmath133 encloses the spectral subset @xmath134\cap\sigma(h_{q , r}^\theta)$ ] ( intersecting the real axis in @xmath135 and @xmath136 ) with the real numbers @xmath137 being such that @xmath138 and @xmath136 in the gap @xmath124 . the hall conductance associated with the energy spectrum up to the gap @xmath124 is related to the projection @xmath125 via the kubo formula ( linear response theory ) . its value is an integer number @xmath139 ; it is by now well known that @xmath139 is to be thought of as the chern number of a bundle determined by the projection @xmath125 @xcite . any such a spectral projection @xmath125 yields a projection @xmath140 , via the representation @xmath46 in ( [ tildepi ] ) , and thus a vector subbundle @xmath141 . the related ( first ) chern number @xmath142 measures the degree of non triviality of the bundle @xmath143 . the geometric interpretation of the hall conductance is none other than the equality @xmath144 . on the other hand , the chern number @xmath142 obeys a diophantine equation which then provides a tknn - type equation for the conductance @xmath139 . we have the following . * proposition b. * _ for any projection @xmath145 in the algebra @xmath146 there exists a vector bundle @xmath147 s.t . the following duality between chern numbers holds : @xmath148.\ ] ] _ before we sketch the proof of this result we turn to its interpretation in terms of conductances of the generalized harper operators in ( [ eq_001_bis ] ) . as mentioned , if @xmath125 is its spectral projection up to the gap @xmath124 , the associated hall conductance @xmath139 is the number @xmath142 . for the dual number we have @xmath149 , with @xmath150 identified with the hall conductance of the energy spectrum up to the gap @xmath124 but in the opposite limit of a weak magnetic field ( @xmath151 ) @xcite . writing @xmath152 , relation ( [ geom_tknn ] ) translates to the _ generalized _ tknn-_equations _ @xmath153 when @xmath154 and @xmath155 , the above reduces to @xmath156 which is the original tknn - equation derived in @xcite for the harper operator ( [ eq_001 ] ) . in its spirit then , the integer @xmath157 in the right - hand side coincides with the labeling of the gap when @xmath45 is odd , i.e. @xmath158 for @xmath45 odd . when @xmath45 is even @xmath158 if @xmath159 and @xmath160 if @xmath161 . we remark that the bound @xmath162 ( already present in @xcite ) still holds , owing to the bound @xmath163 for the spectral projections into the gaps of the hofstadter operator @xcite . now , the subbundle @xmath164 determined by the projection valued section @xmath165 , for a projection @xmath166 , will have as fiber over @xmath167 the space @xmath168 we need a dual bundle representation , @xmath169 of @xmath146 , s.t . @xmath170 and @xmath171 . we are lead to the representation @xmath172 which obey @xmath173 . as elements in @xmath174 they yield a representation of @xmath146 as endomorphisms of the the trivial bundle @xmath175 . then , any projection @xmath145 in @xmath146 is mapped to a projection - valued section @xmath176 which defines a vector subbundle @xmath177 of the trivial vector bundle @xmath178 . it will have as fiber over @xmath167 the space @xmath179 then , equation ( [ eq_proofgd4 ] ) say that the vector bundle @xmath180 around @xmath45 times in the second direction is ( locally ) isomorphic to the vector bundle @xmath181 around @xmath182 times in the same direction . there is however an extra twist , due to the bundle @xmath183 , of which @xmath184 is a subbundle , being not trivial . indeed , an analysis of the transition functions lead to the bundle isomorphism @xmath185 here @xmath186 is the determinant line bundle and the extra operation @xmath187 ( the pullback ) stays for the extra winding by @xmath45 ( for the bundle @xmath180 ) and the same for @xmath188 ( for the bundle @xmath181 ) . formula ( [ geom_tknn ] ) is the relation among corresponding first chern numbers . using the fact that @xmath189 and @xmath190 , as well as the identity @xmath191 , the relation ( [ geom_tknn ] ) follows from ( [ rvb ] ) by standard arguments . on the algebra @xmath31 there is a faithful trace defined by @xmath192 on monomials , and extended by linearity . derivations @xmath193 , for @xmath194 , defined on monomials by @xmath195 are extended by linearity and leibniz rule . lastly , we need the first connes - chern number which , for a projection @xmath196 ( in the domain of the derivations ) computes the integer ( an index of a fredholm operator ) @xmath197 let @xmath198 be the hofstadter operator ( [ eq_001_bis ] ) with associated spectral projection @xmath199 for the gap @xmath124 as in ( [ rf ] ) . for @xmath200 , the functional expression of @xmath198 is fixed and @xmath120 depends on the parameter @xmath201 only through the fundamental commutation relation which defines @xmath31 . now , if the gap @xmath124 is open for all @xmath202 ( with @xmath203 sufficiently small ) , the functions @xmath204 is constant in the interval @xmath203 @xcite . on the other hand , from the structure of the group @xmath205 , one deduces @xcite that @xmath206 with the integer @xmath207 uniquely determined by the condition @xmath208 . from ( [ eq_c10b - bis ] ) , the integer @xmath209 is constant for @xmath202 . hence , formula @xmath210 = q\left[\tau(p_g)+\left(\theta-\frac{r}{q}\right ) { \mathrel{{\ooalign{$c_1$\cr\kern-.13em\raise.45ex\hbox{$\boldsymbol{\scriptstyle-}$}\cr}}}\!\!}(p_g)\right]\in{\mathbb z}\ ] ] is well defined and extends ( [ geom_tknn ] ) for irrational values @xmath202 ( for which the gap @xmath124 remains open ) . indeed , for a rational @xmath43 one has natural identifications @xmath211 and @xmath212 @xcite and for the rational torus , formula ( [ eq_11 ] ) is the same as ( [ geom_tknn ] ) . we think of ( [ eq_11 ] ) as relating conductances for the harper operator @xmath120 in ( [ eq_001_bis ] ) , thus generalizing ( [ gen - tknn - eqs ] ) to @xmath213 with @xmath125 once again the spectral projections of the harper operator @xmath120 , and now identifying @xmath214 and @xmath215 as before , whereas @xmath216 . is supported by the grant anr-08-blan-0261 - 01 . was partially supported by the italian project 08 noncommutative geometry , quantum groups and applications . gl thanks noureddine mebarki and jamal mimouni for the kind invitation to the 8th international conference on progress in theoretical physics held at mentouri university , constantine , algeria , october 2011 . he is grateful to them and to all participants for the great time in constantine .	a generalized version of the tknn - equations computing hall conductances for generalized dirac - like harper operators is derived . geometrically these equations relate chern numbers of suitable ( dual ) bundles naturally associated to spectral projections of the operators . address = dpartement de mathmatiques , universit de cergy - pontoise , 95302 cergy - pontoise , france , [email protected] address = dipartimento di matematica , universit di trieste , i-34127 trieste , italy + and infn , sezione di trieste italy , [email protected]
ultracold neutral atoms in an optical lattice with single - atom and single - site resolution constitute an ideal physical system to investigate strongly correlated quantum phases @xcite which , in turn , has interesting applications in quantum optics @xcite , quantum simulation @xcite and quantum information processing @xcite , among others . the first approaches towards single - site addressing considered the use of lattices with relatively large site separations @xcite . however , to have access to the regime of strongly correlated systems , typical lattice spacings well below @xmath2 m are needed since the tunneling rate has to be comparable to the on site interactions . in this case , the diffraction limit imposes strong restrictions on the addressability of individual lattice sites . to overcome this limitation different techniques have been investigated . for instance , spatially dependent electric and magnetic fields have been used to induce position dependent energy shifts on the atom @xcite , allowing for site - selective addressability . alternatively , a scanning electron microscopy system to remove atoms from individual sites with a focused electron beam @xcite has been reported . however , in this case , atoms need to be reloaded into the emptied sites after each detection event . more recently , high resolution fluorescence imaging techniques , that make use of an optical system with high numerical aperture , have been implemented to perform _ in situ _ single - atom and single - site imaging for strongly correlated systems @xcite . in this context , a single - site addressing ( ssa ) scheme based on focused laser beams inducing position - dependent energy shifts of hyperfine states has been theoretically @xcite and experimentally reported @xcite . in the experiment , an intense addressing beam is tightly focused by means of a high resolution optical system . this beam produces spatial dependent light shifts bringing the addressed atom into resonance with a chirped microwave pulse and eventually inducing a spin - flip between two different hyperfine levels of the atom . on the other hand , during past years , several proposals based on the interaction of spatially dependent fields , _ e.g. _ , standing waves , with three - level atoms in a @xmath0-type configuration have been considered , not only for single - site addressing in optical lattices but , more generally , for sub - wavelength resolution and localization @xcite . in the first approaches @xcite , a spatially modulated dark state is created by means of either electromagnetically induced transparency ( eit ) or coherent population trapping ( cpt ) @xcite , which allows for a tight localization of the atomic population in one of the ground states , around the position of the nodes of the spatially dependent field . more recently , it has been shown that the resolution achieved with those cpt or eit based techniques can be surpassed using stimulated raman adiabatic passage ( stirap ) @xcite processes , by means of the so - called sub - wavelength localization via adiabatic passage ( slap ) technique @xcite . the slap technique relies on a position - dependent stirap of atoms between the two ground states of a three - level @xmath0 atomic system , and has additional advantages compared with cpt or eit techniques such as ( i ) robustness against parameter variations , ( ii ) coherence of the transfer process , that allows for its implementation also in bose - einstein condensates @xcite , and ( iii ) the absence of photon induced recoil . in this work , we apply the slap technique @xcite to ultracold atoms in an optical lattice , where single - site addressing ( ssa ) requires one to overcome the diffraction limit . in order to address only a single site , we use here stokes and pump pulses with gaussian shaped spatial distributions , with the pump presenting a node centered at the lattice site that we want to address . assuming that all the atoms in the optical lattice are initially in the same internal ground state and applying the standard stirap counterintuitive temporal sequence for the light pulses @xcite , we will demonstrate that it is possible to adiabatically transfer all the atoms , except the one at the node of the pump field , to an auxiliary ground state . we will show that this process is performed with higher efficiency and yields better spatial resolution than the cpt based techniques @xcite . also , we will demonstrate that our addressing technique requires shorter times than in the adiabatic spin - flip technique discussed in ref . @xcite , and that larger addressing resolutions can be achieved using similar focusing of the addressing fields . moreover , our technique has the additional advantage that it can be applied between two degenerated ground - state levels . the paper is organized as follows . in sec . [ sec : physicalmodel ] we introduce the physical system under consideration . in sec . [ sec : singlesiteaddressing ] , we present a protocol to achieve ssa , and we derive analytical expressions for the spatial resolution and addressing efficiency of our technique . in addition , a comparison with cpt based techniques is provided . next , in sec . [ sec : numerical simulations ] , we perform a numerical investigation of the proposed technique for a single - occupancy optical lattice loaded with @xmath1rb atoms by integrating the corresponding atomic density - matrix equations . finally , in sec . [ sec : conclusions ] , we summarize the results and present the conclusions . figure [ f : fig1ab](a ) illustrates the physical system under consideration . it consists of a sample of ultracold neutral atoms loaded into a two - dimensional ( 2d ) square optical lattice with spatial period @xmath3 , placed in the plane ( @xmath4 ) and illuminated by a pump and a stokes laser pulse with rabi frequencies @xmath5 and @xmath6 , respectively , propagating in the @xmath7 direction with a selectable time delay . the spatial profile of the pump pulse has a node coinciding with a particular lattice site , our target site , at which the stokes pulse is also centered . with the spatial profiles of the pulses having revolution symmetry around the propagation axis [ dashed line in fig . [ f : fig1ab](a ) ] , in the following we consider only the transverse spatial dimension @xmath8 , without loss of generality . in our model , we assume the system to be in the mott insulator regime with only one atom per lattice site . each atom is considered to have only three relevant energy levels in a @xmath0-type configuration , defined by the interaction with the light pulses as shown in fig . [ f : fig1ab](b ) . here , @xmath9 ( @xmath10 ) is the spontaneous transition rate from the excited state @xmath11 to the ground state @xmath12 ( @xmath13 ) and @xmath14 ( @xmath15 ) is the detuning of the pump ( stokes ) field . we assume that all atoms are initially in state @xmath12 . and @xmath6 , propagate in the @xmath7 direction and interact with the atoms of a single - occupancy optical lattice located in the ( @xmath4 ) plane . ( b ) scheme of the @xmath0-type three - level atoms , initially in state @xmath12 , that interact with pump and stokes pulses . excited level @xmath11 has spontaneous transition rate @xmath9 ( @xmath10 ) to level @xmath12 ( @xmath13 ) and @xmath14 ( @xmath15 ) is the detuning of the pump ( stokes ) field.,title="fig : " ] our approach to achieve single site addressing is based on the slap technique @xcite , where , depending on their position , the atoms are transferred between two internal ground states by means of the stirap technique @xcite . stirap consists in adiabatically following one of the energy eigenstates of the @xmath0 system , the so - called dark state , which under the two - photon resonance condition , _ i.e. _ , @xmath16 , has the form @xmath17 where @xmath18 . starting with all the population in @xmath12 , it is possible to coherently transfer the atomic population to state @xmath13 changing adiabatically @xmath19 from @xmath20 to @xmath21 by means of a convenient time sequence of the fields . this time sequence corresponds to apply first the stokes pulse and , with a certain temporal overlap , the pump pulse . since the process involves one of the eigenstates of the system , the population transfer is robust under fluctuations of the parameter values if these are adiabatically changed and the system does not evolve near degenerate energy eigenvalues . in the slap technique , the pump field has a spatial structure with nodes yielding state - selective localization at those positions where the adiabatic passage process does not occur , _ i.e. _ , those atoms placed at the nodes of the pump field remain in @xmath12 , while those interacting with both fields , pump and stokes , are transferred to @xmath13 . for our purposes , we use the slap technique with a pump field having a single node at the position of the target site . therefore , at the end of the slap process the population of all atoms illuminated is transferred from @xmath12 to @xmath13 except for the one at the node of the pump field . the spatial and temporal profiles for pump and stokes rabi frequencies are given by @xmath22 where @xmath23 and @xmath24 are the peak rabi frequencies , @xmath25 and @xmath26 are the centers of the temporal gaussian profiles , @xmath27 and @xmath28 are the spatial widths of the node in the pump and of the stokes field , respectively , and @xmath29 is the temporal width . there exist several methods to create the required pump intensity profile with a central node : e.g. ( i ) re - imaging of a gaussian beam with a dark central spot created by a circular absorption mask , using ( ii ) a laguerre - gaussian laser beam @xcite or ( iii ) a `` bottle beam '' created by the interferometric overlap of two gaussian beams with differing waists @xcite , or ( iv ) a flexible intensity pattern generated by spatial light modulators and subsequent imaging @xcite . in our model , we assume that the spatial wavefunctions of the individual atoms placed at the different sites , centered at @xmath30 ( being @xmath31 the site index ) , correspond to the ground state of the trapping potential , which in first approximation can be considered harmonic . therefore , the full atomic distribution in the lattice is given , initially , by @xmath32},\end{aligned}\ ] ] where @xmath33 is the width of the initial atomic distribution at an individual site , @xmath34 is the mass of the trapped atom and @xmath35 is the harmonic trapping frequency . we assume that the addressed site is @xmath36 and their nearest neighbors @xmath37 are at a distance @xmath38 where @xmath39 is the wavelength of the fields that create the optical lattice . in order to characterize our single - site addressing technique we consider that , once the slap technique has been applied , the final atomic population distribution in @xmath12 , @xmath40 , is given by @xmath41 where @xmath42 is the probability distribution that an atom remains in state @xmath12 after the slap process . using the slap technique , the addressing resolution that one can obtain is related to the global adiabaticity condition @xcite at each spatial position @xmath8 , @xmath43 ^ 2\geq\left(\frac{a}{t}\right)^2 , \notag \\\end{aligned}\ ] ] where @xmath44 and @xmath45 is a dimensionless constant that , for optimal gaussian temporal profiles and overlapping times , takes values around 10 @xcite . in eq . ( [ eq : adiabatic condition ] ) , the equality gives a spatial threshold @xmath46 above which the adiabaticity condition is fulfilled . assuming that the full width at half maximum ( fwhm ) of @xmath42 is @xmath47 and expanding eq . ( [ eq : adiabatic condition ] ) up to first order in @xmath8 one obtains @xmath48 where @xmath49 and @xmath50 . equation ( [ eq : fwhm prob slap ] ) gives the width of the addressing region , and it tends to zero as @xmath51 increases . moreover , since @xmath52 must be real valued , we find that the inequality @xmath53 must be fulfilled . in this paper we will consider that @xmath24 is fixed , so @xmath54 can be varied through @xmath55 , and @xmath28 . two conditions should be satisfied for our ssa technique to work . first , the population of the atom in the addressed site must remain in state @xmath12 after the action of the fields and , second , the rest of the atoms of the lattice have to be transferred to level @xmath13 . therefore , taking into account the overlap between @xmath42 and @xmath56 in eq . ( [ eq : final distribution ] ) , it is clear that the fwhm of the probability distribution @xmath42 should satisfy @xmath57 where @xmath58 , and @xmath59 is the position of the nearest - neighboring site . using eq . ( [ eq : fwhm prob slap ] ) , it is easy to see that these conditions fix the range for @xmath60 to obtain ssa using the slap technique : @xmath61 where @xmath62 ^ 2+r'}},\end{aligned}\ ] ] with @xmath63 and @xmath64 . note that the upper limit for eq . ( [ eq : ssa range ] ) is more restrictive than eq . ( [ eq : math limit ] ) . in order to have a quantitative description of the ssa performance , let us introduce the ssa efficiency as @xmath65 where @xmath66 corresponds to the probability of finding the atom at the addressed site @xmath67 in state @xmath12 , while @xmath68 corresponds to the probability that the atom in the neighbor site @xmath69 has been transferred to a different internal state . we define @xmath70 with @xmath71 and @xmath72 , whereas @xmath40 and @xmath56 have been defined in eqs . ( [ eq : final distribution ] ) and ( [ eq : wavefunction ] ) , respectively . using eqs . ( [ eq : wavefunction ] ) and ( [ eq : final distribution ] ) , the explicit forms for eqs . ( [ eq : partialefficiency ] ) are @xmath73}. \label{eq : explicitefficiencyx1}\end{aligned}\ ] ] from these expressions , it can be seen that , for @xmath74 , the limits given by eq . ( [ eq : deltax ssa range ] ) , _ i.e. _ , @xmath75 and @xmath76 , correspond to ssa efficiencies of @xmath770.70 and @xmath770.94 , respectively . an alternative technique to perform atomic localization based on spatial dependent dark states is the coherent population trapping ( cpt ) technique @xcite . in the cpt @xcite technique the dark state is populated after several cycles of coherent excitation followed by spontaneous emission from @xmath11 to the ground states . note that , while cpt relies on spontaneous emission , the slap technique is fully coherent . moreover , the latter provides higher resolution , as shown in ref . @xcite , and does not suffer from recoil since the localized atoms have not interacted with light . in what follows , we compare the range of parameters necessary to perform ssa considering both slap and cpt techniques . note that we focus our comparative analysis in the cpt technique , although similar results are obtained considering the method proposed in ref . @xcite , where the spatial dependent dark state , @xmath78 , is created via the stirap technique by switching off the fields before completing the transfer process . in order to compare both techniques , we define the final population distribution in @xmath12 using cpt as @xmath79 in an analogous way as it has been done for the slap technique in eq . ( [ eq : final distribution ] ) . then the fwhm of the corresponding probability function @xmath80 is obtained by imposing that @xmath81 and @xmath82 in eqs . ( [ eq : fieldp ] ) and ( [ eq : fields ] ) : @xmath83 fixing the desired @xmath84 and using eqs . ( [ eq : fwhm prob slap ] ) and ( [ eq : fwhm prob cpt ] ) for slap and cpt , respectively , the constraints for the relevant parameters for each technique can be obtained . for simplicity , we consider that the stokes pulse parameters @xmath28 and @xmath24 are fixed , and only the node width @xmath27 and pump peak rabi frequency @xmath23 can be varied . note that , for the slap case , we have to fix also @xmath45 and @xmath85 . needed to perform ssa as a function of @xmath27 , using slap ( solid line ) and cpt ( dashed line ) techniques . the addressing probability distribution widths are taken as @xmath86 in both cases . ( b ) @xmath87 ( solid line ) and @xmath88 ( dashed line ) as a function of @xmath51 . the fwhm of the atomic distribution , @xmath89 corresponds to the horizontal dotted line and we have taken @xmath90 . the parameters used in both ( a ) and ( b ) are @xmath91 , @xmath92 , and @xmath93 . , title="fig : " ] taking @xmath86 , half of the site separation , the required values for @xmath51 and @xmath27 are plotted in fig . [ f : fig2](a ) . to simultaneously illuminate a large number of sites , we use a large stokes beam waist of @xmath91 , where @xmath94 is the wavelength of both pump and stokes fields @xcite . the solid ( dashed ) line corresponds to the slap ( cpt ) case with the parameter values @xmath92 and @xmath93 @xcite . as @xmath27 decreases , in both slap and cpt cases , lower values of @xmath51 are needed to reach the fixed resolution @xmath84 , since the narrower the node of the pump field , the narrower the probability distribution of atoms remaining in @xmath12 . in addition , for any given width of the node of the pump field @xmath27 , the required values of @xmath51 are lower in the slap case than in the cpt case . it is important to realize that eqs . ( [ eq : fwhm prob slap ] ) and ( [ eq : fwhm prob cpt ] ) show the possibility to obtain , for certain parameter values , widths of the probability distribution , @xmath87 or @xmath88 , smaller than @xmath95 . in particular , using the slap technique this can be achieved with moderate @xmath51 values . this is shown in fig . [ f : fig2](b ) , where @xmath87 ( solid line ) and @xmath88 ( dashed line ) are represented as a function of @xmath51 for @xmath96 and the rest of the parameters as in fig . [ f : fig2](a ) . the fwhm of the atomic distribution , @xmath89 , is depicted with a horizontal dotted line to indicate the values where @xmath97 . as we stated in the discussion of eqs . ( [ eq : explicitefficiencyx0 ] ) and ( [ eq : explicitefficiencyx1 ] ) , this limit corresponds to a ssa efficiency of @xmath98 . this regime of parameters is interesting because it shows that the slap technique could be used for applications in site - selective imaging with a resolution down to the width of the atomic distribution at each site . in this section , by numerically integrating the corresponding density - matrix equations , we study the implementation of the slap - based ssa technique for @xmath0-type three - level @xmath1rb atoms in a single - occupancy optical lattice . numerical calculations using the cpt technique are also presented for comparison . the wavelength of the lasers that create the optical lattice , red detuned with respect to the @xmath99 line of @xmath1rb , is @xmath100 nm . the potential depth of the optical lattice is chosen as @xmath101 , where @xmath102 is the recoil energy . this corresponds to a harmonic trapping frequency of @xmath103 khz @xcite . therefore , the fwhm of the atom distribution at each site due to the confining potential is @xmath104 nm . pump and stokes fields with @xmath105 nm are coupled to @xmath106 and @xmath107 , respectively , where @xmath108 , @xmath109 and @xmath110 are hyperfine energy levels of the @xmath99 line of @xmath1rb . the excited state @xmath11 has a spontaneous transition rate @xmath111 mhz ( @xmath112 mhz ) to state @xmath12 ( @xmath13 ) , and we assume no spin decoherence during the interaction time . we consider equal temporal pulse widths of @xmath113s with a temporal delay @xmath114 , in such a way that the total ssa process time is @xmath115 . the stokes pulse has a maximum rabi frequency @xmath116 mhz , while the maximum rabi frequency of the pump is varied through the parameter @xmath51 , since @xmath117 . concerning the spatial profiles of the fields , we assume a wide stokes profile , @xmath118 , and a narrow node for the pump , @xmath90 . in state @xmath12 ( a ) , the probability to transfer it from @xmath12 to another state ( b ) , and the efficiency @xmath119 as a function of @xmath51 ( c ) , for the ssa with slap ( circles ) and cpt ( crosses ) techniques . analytical curves for slap ( solid line ) and cpt ( dashed lines ) , computed from eqs . ( [ eq : totalefficiency ] ) and ( [ eq : partialefficiency ] ) , are added in ( c ) for comparison ( see text for the rest of the parameters).,title="fig : " ] as it has been discussed in the previous section , to properly perform ssa , the population of all the atoms in the lattice , except the one in the addressed site , must be transferred from @xmath12 to @xmath13 with high probability . in what follows , those requirements for the realization of the ssa are studied by numerically evaluating the ssa efficiency . the signatures of ssa are shown in fig . [ f : efficienciesnum ] , where the numerically evaluated efficiency and probabilities defined in eqs . ( [ eq : totalefficiency ] ) and ( [ eq : partialefficiency ] ) are plotted as a function of @xmath51 for both the slap ( circles ) and cpt ( crosses ) techniques . figures [ f : efficienciesnum](a ) and [ f : efficienciesnum](b ) show the probabilities @xmath120 and @xmath121 , respectively . for large values of @xmath51 the probability of finding the atom at the addressed site in @xmath12 is higher with the cpt than with the slap technique [ see fig . [ f : efficienciesnum](a ) ] , while for small values of @xmath51 the probability of removing the atom from the neighboring site [ see fig . [ f : efficienciesnum](b ) ] is higher in the slap case . this is explained by the fact that @xmath122 for a given value of @xmath51 , as shown in fig . [ f : fig2](b ) . the ssa efficiency results obtained by the numerical evaluation of eqs . ( [ eq : totalefficiency ] ) and ( [ eq : partialefficiency ] ) are shown in fig . [ f : efficienciesnum](c ) , together with the corresponding analytical curves ( solid and dashed lines ) obtained using eqs . ( [ eq : explicitefficiencyx0 ] ) and ( [ eq : explicitefficiencyx1 ] ) , added for comparison . a good agreement is found between numerical and analytical results . from fig . [ f : efficienciesnum](c ) it is clear that the slap technique is more efficient for lower values of the intensity ratio of the addressing fields ( @xmath123 ) than the cpt technique . certainly , this is advantageous for the experimental implementation with limited laser power available . . ( b ) final population distribution remaining in @xmath12 using slap ( circles ) and cpt ( crosses ) single - site addressing techniques for @xmath124 . the initial atomic distribution in @xmath12 , @xmath56 , is shown as a solid line ( see text for the parameters values ) . , title="fig : " ] an example of the final population distribution after performing ssa with the slap ( circles ) and the cpt ( crosses ) techniques is plotted in fig . [ f : rubidiumpeaks](b ) for the particular case of @xmath124 and with the rest of the parameters as in fig . [ f : efficienciesnum ] . the initial population distribution in @xmath12 , @xmath56 , at the addressed site ( @xmath125 ) , and the two next neighbors @xmath126 nm is shown as a solid line . note that , in the slap case ( circles ) , the population of state @xmath12 around @xmath67 remains almost the same after the addressing process , while in the first neighbor sites it is practically zero . on the other hand , for the cpt case ( crosses ) , the population in the addressed site remains also nearly unchanged , but it exhibits a significant amount of population in the neighbor sites . this is in full agreement with the discussion following figs . [ f : efficienciesnum](a ) and [ f : efficienciesnum](b ) . for this example , the total efficiencies found are @xmath127 and @xmath128 according to the corresponding values shown in fig . [ f : efficienciesnum](c ) . in addition , as shown in fig . [ f : rubidiumpeaks](a ) , the width of the pump node required to perform the ssa method is much larger than the addressed region @xmath87 . in particular , for the case shown in fig . [ f : rubidiumpeaks ] , @xmath129 and @xmath130 , thus obtaining addressing resolution beyond the diffraction limit . finally , we carry out a comparison between our proposal and the experiment reported in ref . @xcite , where a focused laser beam induces position - dependent light shifts , allowing one to perform a spin flip by means of a resonant microwave pulse at the addressed site . since the microwave field involved in the experiment has a rabi frequency of khz , the total spin - flip time is in the order of ms . in contrast , as our proposal makes use of only optical fields , the addressing time is three orders of magnitude below ( @xmath131s ) . specifically , to achieve similar values of the addressing resolution in both techniques , @xmath132 nm , we have obtained an addressing time of @xmath7740@xmath133s . this decrease of the total addressing time needed implies a reduction of the effects caused by spontaneous scattering of photons , which are a limitation for the light shifts - based proposals @xcite . these effects could be strongly reduced in our case . also , we have compared the resolution obtained in both techniques . in ref . @xcite , they use an addressing beam with an intensity fwhm of approximately 600 nm , and obtain a spin - flip probability distribution with fwhm=330 nm . in our technique , using @xmath134 and a width of the pump node @xmath135 nm , which corresponds to the width of their addressing beam , a very similar fwhm of the addressing probability distribution is obtained : @xmath136 nm . note that this value can be reduced by increasing the ratio between the pump and the stokes intensities , _ e.g. _ , @xmath137 implies @xmath138 nm and for @xmath139 we obtain @xmath140 nm . in this work we have discussed a proposal to perform single - site addressing ( ssa ) in an optical lattice by using the slap technique . with respect to other dark state based techniques such as cpt , this method is fully coherent , robust against variations of the parameter values , and we have found that it yields higher efficiencies for smaller values of the intensity ratio between the pump and the stokes fields . moreover , the addressed atom does not interact with the fields , minimizing all possible decoherent processes . on the other hand , with respect to the recent experiment on ssa using adiabatic spin flips @xcite , the present proposal allows one to use two degenerate ground levels , takes shorter times to perform the addressing process , and provides similar or even larger addressing resolutions for similar focusing of the addressing fields . the proposed method provides an achievable addressing resolution that can be pushed well below the diffraction limit of the addressing light field and of the optical setup used for addressing or detection of atoms at closely spaced lattice sites . this relaxes the requirements on the optical setup or extends the achievable spatial resolution to lattice spacings smaller than accessible to date . through analytical considerations , we have derived the range of parameters for which ssa is properly achieved . moreover , we have obtained expressions to estimate the resolution and the efficiency of the slap - based addressing method , and we have compared them with the analogous expressions obtained using the cpt technique . next , by integrating the density - matrix equations with realistic parameter values for state - of - the - art optical lattices loaded with @xmath1rb atoms , we have checked the validity of the analytical approach . we thank albert benseny for fruitful discussions and comments . we acknowledge financial support from the spanish ministry of science and innovation under contracts no . fis2008 - 02425 , no . fis2011 - 23719 , and no . csd2006 - 00019 ( consolider project quantum optical information technologies ) , from the catalan government under contract no . sgr2009 - 00347 , from the german research foundation dfg ( contract no . bi 647/3 - 1 ) , and from the german academic exchange service daad ( contract no . 0804149 ) . 99 m. endres , m. cheneau , t. fukuhara , c. weitenberg , p. schau , c. gross , l. mazza , m. c. banuls , l. pollet , i. bloch and s. kuh , science * 334 * , 200 ( 2011 ) . n. s. ginsberg , s. r. garner , and l. v. hau , nature * 445 * , 623 ( 2007 ) . i. bloch , j. dalibard , and w. zwerger , rev . phys . * 80 * , 885 ( 2008 ) . g. k. brennen , c. m. caves , p. s. jessen , and i. h. deutsch , phys . lett . * 82 * , 1060 ( 1999 ) . r. raussendorf and h. j. briegel , phys . rev . lett . * 86 * , 5188 ( 2001 ) . d. jaksch and p. zoller , ann . 315 , * 52 * ( 2005 ) . r. scheunemann , f. s. cataliotti , t. w. hnsch , and m. weitz , phys . a * 62 * , 051801(r ) ( 2000 ) ; r. dumke , m. volk , t. mther , f. b. j. buchkremer , g. birkl , and w. ertmer , phys . * 89 * , 097903 ( 2002 ) ; s. peil , j. v. porto , b. laburthe tolra , j. m. obrecht , b. e. king , m. subbotin , s. l. rolston , and w. d. phillips , phys . rev . a * 67 * , 051603(r ) ( 2003 ) ; y. miroshnychenko , w. alt , i. dotsenko , l. frster , m. khudaverdyan , d. meschede , d. schrader , and a. rauschenbeutel , nature 442 , * 151 * ( 2006 ) ; k. nelson , x. li , and d. weiss , nature phys . * 3 * , 556 ( 2007 ) ; g. birkl and j. fortgh , laser & photon . rev . * 1 * , 12 ( 2007 ) . w. s. bakr , j. i. gillen , a. peng , s. flling , and m. greiner , nature * 462 * , 74 ( 2009 ) ; w. s. bakr , a. peng , m. e. tai , r. ma , j. simon , j. i. gillen , s. flling , l. pollet , and m. greiner , science * 329 * , 547 ( 2010 ) ; n. gemelke , x. zhang , c .- l . hung , and c. chin , nature * 460 * , 996 ( 2009 ) . m. karski , l. frster , j. m. choi , w. alt , a. widera , and d. meschede , phys . * 102 * , 053001 ( 2009 ) ; j. f. sherson , c. weitenberg , m. endres , m. cheneau , i. bloch , and s. kuhr , nature * 467 * , 68 ( 2010 ) . g. s. agarwal , and k. t. kapale , j. phys . b : at . mol . . phys . * 39 * , 3437 ( 2006 ) . j. cho , phys . * 99 * , 020502 ( 2007 ) ; a. v. gorshkov , l. jiang , m. greiner , p. zoller , and m. d. lukin , phys . lett . * 100 * , 093005 ( 2008 ) ; m. holland , s. marksteiner , p. marte , and p. zoller , phys . rev . lett . * 76 * , 3683 ( 1996 ) ; e. paspalakis and p. l. knight , phys . rev . a * 63 * , 065802 ( 2001 ) ; m. sahrai , h. tajalli , k. t. kapale , and m. s. zubairy , phys . a * 72 * , 013820 ( 2005 ) ; m. kiffner , j. evers , and m. s. zubairy , phys . lett . * 100 * , 073602 ( 2008 ) . d. d. yavuz and n. a. proite , phys . rev . a * 76 * , 041802(r ) ( 2007 ) . j. mompart , v. ahufinger , and g. birkl , phys . rev . a * 79 * , 053638 ( 2009 ) . e. arimondo , in _ progress in optics _ , edited by e. wolf ( elseveir science , amsterdam , 1996 ) , xxxv , p. 257 . g. alzetta , l. moi and g. orriols , nuovo cimento * 52b * , 209 ( 1979 ) ; s. e. harris , phys . today * 50(7 ) 36 , ( 1997 ) ; j. p. marangos , j. mod . opt . 45 * , 471 ( 1998 ) . heckenberg , r. mcduff , c.p . smith , h. rubinstein - dunlop , and m.j . wegener , opt . quantum electron . * 24 * , s951 , 1992 . l. isenhower , w. williams , a. dally , and m. saffman , optics letters * 34 * , 1159 ( 2009 ) . s. bergamini , b. darquie , m. jones , l. jacubowiez , a. browaeys , and p. grangier , j. opt . b * 21 * , 1889 , 2004 . j. kruse , c. gierl , m. schlosser , and g. birkl , phys . a * 81 * , 060308(r ) , 2010 .	we propose a single - site addressing implementation based on the sub - wavelength localization via adiabatic passage ( slap ) technique . we consider a sample of ultracold neutral atoms loaded into a two - dimensional optical lattice with one atom per site . each atom is modeled by a three - level @xmath0 system in interaction with a pump and a stokes laser pulse . using a pump field with a node in its spatial profile , the atoms at all sites are transferred from one ground state of the system to the other via stimulated raman adiabatic passage , except the one at the position of the node that remains in the initial ground state . this technique allows for the preparation , manipulation , and detection of atoms with a spatial resolution better than the diffraction limit , which either relaxes the requirements on the optical setup used or extends the achievable spatial resolution to lattice spacings smaller than accessible to date . in comparison to techniques based on coherent population trapping , slap gives a higher addressing resolution and has additional advantages such as robustness against parameter variations , coherence of the transfer process , and the absence of photon induced recoil . additionally , the advantages of our proposal with respect to adiabatic spin - flip techniques are highlighted . analytic expressions for the achievable addressing resolution and efficiency are derived and compared to numerical simulations for @xmath1rb atoms in state - of - the - art optical lattices .
it has been known since gauss that every complex polynomial @xmath3 in a single complex variable splits into linear factors , and since ruffini and abel that there is no method based on iterated @xmath4-th roots to find these factors algebraically in general . therefore , numerical approximation methods are required to find the roots of polynomials . ( detail ) . the colors illustrate the number of iterations until a root is found to high precision ; adjacent color contour lines show a single iteration . the scale is between 50 iterations ( red ) and 0 iterations ( dark blue ) ; bright green stands for 25 iterations . compare with figure [ fig : newtonmarvin_1_24 ] . picture by marvin randig . , scaledwidth=60.0% ] surprisingly , even today it is not clear how to find all roots most efficiently . there is a theoretically best possible algorithm due to pan @xcite , but it is not applicable in practice . perhaps the most efficient practical implementation is ` mpsolve 3.0 ` by bini and robol @xcite based on the ehrlich - aberth iteration , but it is still lacking good theory and especially a proof that it is generally convergent . newton s method is very good at finding roots locally , i.e. once good approximations are known , but it has a reputation of being difficult to predict globally , for instance because of the `` chaotic '' nature of the iteration , and because of the possibility of having open sets of starting points that converge to attracting periodic orbits , rather than to roots . the situation has improved substantially in recent times . indeed , a pleasant feature of newton s method is that one does not have to choose between theory and practical implementations : meanwhile it has good theory ( see @xcite on where to start the iteration , and @xcite with refinements in @xcite on estimates on its complexity that are near - optimal under certain assumptions ) , and it can be implemented quite successfully in practice for polynomials of very high degrees . the latter aspect is the key theme of the present manuscript . in @xcite , we showed that newton s method can be used to find all roots of various complex polynomials efficiently , even up to degrees greater than one million ( where the polynomials were selected only on the criterion that they could be evaluated efficiently ) . in this note , we demonstrate that the computational complexity ( measured in computing time ) can be near - optimal for a wide range of degrees : all roots can be found in computing time between @xmath1 and @xmath5 , and in a number of newton iterations between @xmath1 and @xmath6 , with small constants , so that all roots can be found for degrees up to @xmath7 ( greater than @xmath8 million ) , and even higher degrees seem feasible . we present an algorithm based on newton s method that turns out to find all roots , the families of polynomials studied , and we describe the outcome of computer experiments . we also explain how to verify that indeed all roots of our polynomials have been found . this note should be seen as a continuation and improvement on @xcite , not as a survey on root finding for polynomials . there are interesting references on the latter , in particular the survey articles by mcnamee @xcite , as well as studies by pan @xcite , renegar @xcite , and especially mpsolve 3.0 from @xcite and eigensolve from @xcite . * acknowledgements. we are very pleased to be able to report support , encouragement , and helpful suggestions from numerous friends and colleagues , especially dario bini , marcel oliver , victor pan , marvin randig , simon schmitt , and michael stoll . this research was supported in part by the advanced grant `` hologram '' of the european research council ( erc ) . there are a number of theoretical results available on the global dynamics of newton s method @xmath9 for a given polynomial @xmath3 . all require that a disk is known that contains all roots of @xmath3 . after appropriate rescaling , we may assume that all roots of @xmath3 are contained in @xmath10 . we showed in @xcite that there is a universal set of @xmath11 starting points , placed on @xmath12 concentric circles containing @xmath1 points each , so that newton s method started at these points will find all roots of all polynomials of degree @xmath13 . this was improved in @xcite to a probabilistic set of starting points containing only @xmath14 starting points that find all roots with arbitrarily high probability . our earlier experiments in @xcite started with @xmath15 or @xmath16 equidistributed points on a single circle ; in many cases , @xmath15 points were sufficient , but always @xmath16 points were enough : this shows that in practice @xmath17 starting points seem to suffice , even though this is not supported by theory ( and there may well be special polynomials that require more starting points ) . of course , any additional factors like @xmath18 are hard to detect numerically , even when the degree ranges up to @xmath7 . theoretical upper bounds on the expected number of required newton iterations were given in @xcite ; these are , up to polylogarithmic factors , on the order of @xmath19 or even @xmath20 . however , these previous methods all require at least @xmath21 iterations in order to find all roots : the methods rely on controlled newton dynamics away from @xmath10 , so all starting points @xmath22 are uniformly bounded away from the disk : @xmath23 for a uniform @xmath24 . however , on this domain we essentially have @xmath25 ( ? ? ? * lemma 3 ) , so each orbit needs at least @xmath26 iterations until it even enters @xmath10 , and since at least @xmath13 orbits are required to find @xmath13 roots , we obtain a lower bound for the complexity of @xmath21 newton iterations . the upper bounds in @xcite are thus best possible , again up to polylogarithmic factors . one may try to remedy this problem by starting the iteration on a circle of radius , say , @xmath27 . such an approach comes with various problems . one is that we are losing the theory to guarantee that all roots will be found . another one is that this will be helpful only if a very tight bound for the smallest disk containing all roots is known , and only if most roots are near the boundary of that disk . this is not the case for instance for all polynomials considered by us ( but it is true if the coefficients of the polynomial are independently randomly distributed @xcite ) . one key observation is that the lower bound of the number of iterations comes from the iterations outside of @xmath10 , that is before the interesting dynamics even starts , and where all starting points are on very similar `` parallel '' orbits a necessary consequence of the fact that we start the newton dynamics at points with controlled dynamics . typical initial orbits of the newton iteration are shown in figure [ fig : paralleliterations ] . the approach we take in the experiments described in this paper is that on domains where orbits are `` parallel '' , fewer orbits are required , as indicated in figure [ fig : newtonrefinement ] . starting points on a circle of radius @xmath28 ( here for a polynomial of degree @xmath29 , so we do not have enough starting points to find all roots ; the polynomial shown here describes periodic points of period dividing @xmath30 of @xmath31 ) . the apparent lines connect orbits under the newton dynamics ; colors indicate the number of iterations until an approximate root is found . the behavior of the iterations outside of the disk containing all roots is very parallel and `` wasteful '' , but required to carry over the control from the circle of starting points to the interesting dynamics on @xmath10 . observe that even if we had a very precise bound on the smallest disk containing all roots , this would not help much as most roots are away from the boundary of this disk . , scaledwidth=65.0% ] points and refine only when adjacent orbits stop moving in parallel . as a result , most newton iterations are spent near the roots , and far fewer iterations are required . the polynomial is the same as in figure [ fig : paralleliterations ] ( degree @xmath29 ) . , scaledwidth=65.0% ] more precisely , we start the newton dynamics with a fixed number of , say , @xmath32 newton orbits on a circle away from @xmath10 where we have good control . along each triple of adjacent orbits , we compare the shapes of the triangles formed by the three points on adjacent orbits . as long as this shape remains nearly constant , so that the three orbits form nearly similar triangles , we keep iterating these three orbits , assuming that they represent similar dynamics for all orbits that might run between these three `` representative '' orbits . once the triangles start to deform , we split up these orbits by inserting an additional orbit half way between each of the two pairs of adjacent orbits ; see figure [ fig : newtonrefinement ] . of course , this requires a heuristic refinement threshold parameter that describes the threshold of deformation after which parallel orbits are refined . we use the cross - ratio between three adjacent orbits with respect to @xmath33 : that is , @xmath34 where the index @xmath4 counts iterations and the index @xmath35 counts the circular order of the orbits ( with due modifications at the end to turn the linear order of natural numbers into a circular order ) . that is , without refinement , we have @xmath36 ; and when refinements occur , these are inserted into the circular order ( and the indices @xmath35 are shifted accordingly ) . our refinement condition is essentially based on @xmath37 , where @xmath38 is the most recent iteration when a refinement occurred for this orbit ( properly accounting for index shifts because of refinements that might have occurred for other orbits ) . the algorithm is determined by a number of parameters that we determined heuristically : * the number @xmath39 of initial orbits ( we usually use @xmath40 , but this should not make much of a difference ) ; * the maximal number of iterations a particular orbit can run for ( we use @xmath41 ) ; more important than this number is a good detection of periodic cycles ( if a positive fraction of orbits runs to the maximal number of allowed iterations , we end up at complexity @xmath21 ) ; * the maximal number of refinement generations ( we use @xmath42 ; that means that if all refinements have taken place , we end up with @xmath15 orbits ) ; * the refinement threshold @xmath43 : that is , the maximal value of @xmath44 that is allowed before further refinement occurs . most of the time , we use the value of @xmath45 , but for some experiments we had to increase sensitivity to @xmath46 ; * and a stopping criterion when success is declared that some orbit found a root : we stop when @xmath47 ; in practice , we use @xmath48 or sometimes @xmath49 ( while we used the ` long double ` data type with numerical precision of about @xmath50 relevant digits ) . -1 finally , one needs a post - processing step : many roots will be found by several different orbits , and we have to make sure these roots are accounted for only once . here we use very simple - minded heuristics and declare that two orbits found different roots if they terminate at distance greater than @xmath51 ; here we use @xmath52 ( sometimes @xmath53 ) . of course , this requires an efficient way to sort the roots by mutual proximity ( when all roots are found by an algorithm that is linear up to log - factors , then for naive sorting procedures it is easy to spend more time on sorting than on root finding ) . of course all quantities are heuristic , and especially the choice of stopping criterion and post - processing ( distinguishing different roots found ) have obvious difficulties in the presence of high degrees or near - multiple roots . our point is not that this is a strong part of the algorithm , but that even with such a simple - minded approach all roots can be found for all the polynomials investigated , and for extremely large degrees . an improved approach for more `` challenging '' families of polynomials , especially with near - multiple roots , is discussed in @xcite . in section [ sec : allrootsfound ] we describe several methods that establish a posteriori that all roots have been found with high accuracy . we investigate ( a subset of ) the same polynomials we already discussed in @xcite . we will briefly introduce the relevant families of polynomials and then present the results of the experiments . these polynomials have been chosen solely for the purpose of fast evaluation ( in terms of recursion ) : our focus is on root finding not efficient polynomial evaluation , so we chose polynomials where evaluation is easy . therefore , comparing our experiments with other experiments on different polynomials in terms of computing time yields an unfair bias in our favor , but comparisons in terms of the required number of newton iterations should be meaningful . these polynomials in a variable @xmath54 are defined by recursion @xmath55 , @xmath56 , so we have @xmath57 , @xmath58 , @xmath59 etc . , so that @xmath60 has degree @xmath61 . roots of @xmath60 are those parameters @xmath54 for which the iteration @xmath62 , @xmath63 is periodic with period ( dividing ) @xmath4 ; this implies that @xmath64 divides @xmath60 when @xmath65 divides @xmath4 ( using the fact that all roots of all @xmath60 are simple ; compare ( * ? ? ? * section 19 ) ) . roots of @xmath60 are known as _ centers of hyperbolic components of the mandelbrot set _ of period @xmath4 ( except those that are already roots of @xmath64 with @xmath66 ) . we found all roots for @xmath67 ( i.e. degree @xmath68 , greater than 16 million ) in less than 160 hours ( on a standard pc using a single core ) . the detailed results are tabulated in figure [ fig : mandelbrotcomplexitytable ] . in particular , we plot the complexity in terms of required newton iterations in figure [ fig : mandelbrotiterationcomplexity ] : the diagram clearly shows that the complexity in terms of newton iterations scales better than @xmath69 . the complexity in terms of of computing time is shown in figure [ fig : mandelbrottimecomplexity ] ; it seems to scale better than @xmath70 . this is of course related to the fact that our degree @xmath13 polynomials can be evaluated in @xmath71 operations . we realize that this is an untypical advantage of our polynomials . however , for polynomials given in coefficient form , there are fast methods of parallel evaluation at many points that should compensate for much of the complexity gain @xcite , ( * ? ? ? * section 8.5 ) ( these methods are efficient only when the number of evaluation points is comparable to the degree of the polynomial , which is the case for the newton algorithm because eventually all roots have to be found by their own newton orbit ) . in order to find all roots , a significantly more sensitive refinement threshold of @xmath46 was used ( compared to @xmath45 used in the other experiments ) . we should point out that newton s method occasionally encounters attracting periodic orbits of periods greater than one . the total number of orbits that were detected to converge to attracting cycles of periods @xmath72 or more is shown in column j of figure [ fig : mandelbrotcomplexitytable ] ; for large @xmath13 , this number seems to stabilize near @xmath73 ( column k and figure [ fig : mandelbrothighercycles ] ) . while the total number of such orbits is relatively small , the orbits do not satisfy the easy - to - detect termination condition that a root is found , so the iteration times can be large . hence , if such orbits are not caught efficiently , they will consume a lot of computing time . it turns out that for the mandelbrot center polynomials , far more attracting cycles were found than in most other experiments . for comparison , for the other families of polynomials investigated here ( periodic points of @xmath74 and @xmath75 ) , for degree @xmath76 no attracting cycles were found at all , and for degree @xmath7 less than one hundred orbits converged to attracting cycles . ( column g ) vs. period . evidently the number of required iterations scales better than @xmath77 . ] ( column i ) . evidently , the time complexity scales better than @xmath70 ( in @xmath78 . ] : for large @xmath13 , this number seems to stabilize near @xmath79 . ] for the quadratic polynomial @xmath80 , a periodic point of period @xmath4 is a root of @xmath81 , where @xmath82 stands for the @xmath4-th iterate of @xmath83 . there are exactly @xmath84 periodic points of period @xmath4 . we found all periodic points of period @xmath85 , i.e. for degree up to @xmath0 million , using a refinement threshold @xmath45 . the largest degree polynomial took about 89 hours . the results are given in detail in figure [ fig : quadratic2complexitytable ] . the complexity in terms of required newton iterations is approximately @xmath86 ( see figure [ fig : quadratic2iterationcomplexity ] ) . the computing time ( shown in figure [ fig : quadratic2timecomplexity ] ) is in the order of @xmath87 for degrees @xmath13 up to 134 million . however , it oscillates significantly by almost a factor of @xmath72 . since we could not detect a clear reason for this oscillation , we ran the experiment twice . the outcome is very similar , except for large periods . these oscillations have no impact on the overall results of our experiments ( but should be investigated in future experiments ) note that this experiment would be outright impossible when expressing @xmath82 in coefficient form : the constant coefficient alone would have magnitude greater than @xmath88 , so for @xmath89 greater than @xmath90 . just storing this number would require about 15 megabytes per coefficient ( note that good relative precision is not sufficient because all the large terms are subtracted eventually ) , and we have to accommodate many million coefficients of possibly different magnitudes . for periods @xmath85 , divided by two scale functions . blue curve : time divided by @xmath91 ( column h in figure [ fig : quadratic2complexitytable ] ) ; red curve : time divided by @xmath92 ( column i ) , multiplied by @xmath93 to adjust the scale . the time complexity seems to scale around @xmath94 ( in @xmath95 , with some fluctuations ) . two different runs of the same experiment are shown in different shades of color ( the number of iterations remained the same).,scaledwidth=75.0% ] for periods @xmath85 , divided by two scale functions . blue curve : time divided by @xmath91 ( column h in figure [ fig : quadratic2complexitytable ] ) ; red curve : time divided by @xmath92 ( column i ) , multiplied by @xmath93 to adjust the scale . the time complexity seems to scale around @xmath94 ( in @xmath95 , with some fluctuations ) . two different runs of the same experiment are shown in different shades of color ( the number of iterations remained the same).,scaledwidth=75.0% ] we performed the same experiments for @xmath96 . the outcome was similar . we found ( almost ) all periodic points for periods up to @xmath89 , again with refinement threshold @xmath45 . this time , the computation was even faster : for period @xmath97 , it took 19 hours ( but 3 roots remained missing ) ; for @xmath98 , all roots were found in 9 hours . for periods @xmath85 . blue : iterations/@xmath99 ( column f in figure [ fig : quadratic_i_complexitytable ] ) ; red : iterations/@xmath100 ( column g ) , multiplied by @xmath93 . the number of iterations seems to grow slightly faster than @xmath101 , but much slower than @xmath102 . , scaledwidth=72.0% ] ) required for finding all periodic points of @xmath75 for periods @xmath85 , divided by three scale functions . blue curve : time divided by @xmath99 ( column h in figure [ fig : quadratic_i_complexitytable ] ) ; red curve : time divided by @xmath91 ( column j ) , multiplied by @xmath103 to adjust the scale ; green curve : time divided by @xmath104 ( column i ) , multiplied by @xmath105 . the time complexity seems to scale with @xmath106.,scaledwidth=80.0% ] one might argue that 3 out of 134 million roots missing are negligible . however , our goal is to find all roots without exception . of course we could have repeated the last experiment with slightly improved sensitivity . instead , we recovered the missing roots by different methods ; see section [ sec : missingroots ] ) . interestingly , one was very near @xmath107 and the other two were very near @xmath108 . the computing time seems to oscillate somewhat unexpectedly in some of our experiments ( see for instance figure [ fig : quadratic2timecomplexity ] ) . a natural explanation might come from the fact that the computers had some other tasks running on separate threads ( perhaps on separate cores ) , and while the actual computations would probably require the same amount of computing time , our memory - intense computations might suffer from unexpected memory swaps to the disk . a re - run of the most suspicious experiments showed of course the same number of newton iterations ( this part is deterministic ) , but also a rather similar oscillation of computing time as before ( also shown in figure [ fig : quadratic2timecomplexity ] ) . a first step to analyze this behavior consists of comparing computing time per number of newton iteration steps . since every newton iteration for our degree @xmath13 polynomials requires complexity @xmath12 , we show in figure [ fig : computingtime_rescaled ] computing time divided by @xmath71 . not surprisingly , this is roughly constant for most series of experiments . interestingly , for the periodic points of @xmath75 one observes a certain decrease in @xmath13 . this might be explained by the fact that not most of the computing time is spent on the newton iterations , but other tasks require a significant fraction of the time . because every evaluation of the polynomial should roughly have complexity @xmath71 ; combined for the three sets of experiments . interestingly , for some experiment these numbers seem to decrease . ( this seems to indicate that the newton iterations do not use most of the computing time , perhaps because the recursion can be implemented particularly efficiently , so other tasks use a definite proportion of the computing time . ) of course , absolute numbers depend on the specific hard- and software used , but the relative behavior should be of structural interest . , scaledwidth=85.0% ] of course , all computing times specified depend on the computer and its software used , in particular efficiency of the compiler and the number of cores available . therefore , the absolute scales are machine dependent and not very relevant ; however , we believe that the relative behavior for changing degrees is of structural interest . the issue of computing time oscillations warrants further study ( but stays within a factor of @xmath72 , often less , and is thus not relevant to the key findings of our experiments ) . we admit that our method is heuristic : we do not have an a - priori guarantee that all roots will be found by the iterated refinement newton algorithm . however , there are several possibilities for testing a posteriori that all roots have indeed been found . several of them have been described in detail in ( * ? ? ? * sections 2.3 and 2.4 ) , so we only briefly mention them here ; an additional method based on the newton identities is described in more detail . an easy observation shows that for any polynomial @xmath3 of degree @xmath13 with associated newton method @xmath109 , and for arbitrary @xmath110 , at least one root is contained in the disk @xmath111 with @xmath112 : so small newton displacement steps @xmath113 occur only near the roots . the factor of @xmath13 can be significantly improved if the approximate locations of many roots are already known ( * ? ? ? * lemma 4 ) . therefore , if we have @xmath13 points ( coming from @xmath13 different newton orbits ) that all come with small disjoint disks containing at least one root , then they together describe all roots . this method worked in practice for all degree @xmath114 polynomials investigated in @xcite , but it requires that there are no near - multiple roots . an independent and simpler test can be performed based on the vite formulas : the coefficients of the polynomial are the elementary symmetric functions of the roots and encode for instance the sum of all roots or their product , and thus provide independent tests of success . more systematically , one can easily compute the value of the power sums @xmath115 over all roots @xmath116 directly from the polynomial @xmath3 ; this approach is based on the well - known newton identities that relate the power sums to the coefficients of @xmath3 . comparing these with the power sums of all roots found is a natural generalization of the vite tests . we describe this in detail below ; the overall results are as follows . in all experiments we describe , we found all roots according to the stopping and distinction criteria described in section [ sec : iteratedrefinementnewton ] , and then used the power sum criterion to verify the roots found . for all our polynomials , even of maximal degrees , the power sum tests for exponents @xmath117 until @xmath118 were run successfully and give reason to conclude that newton s method with our ad - hoc stopping criterion described above ( requiring that @xmath119 ) found all roots with a typical accuracy of no more than @xmath120 . as an example , for the @xmath121 periodic points of @xmath122 of period @xmath97 , the numerically computed power sum for @xmath117 was @xmath123 with a deviation of less than @xmath124 from the true integer value @xmath125 computed via the newton identities . if we assume that all roots were computed with equally and independently distributed errors of expected size @xmath126 , so that the summed error can be viewed as a brownian motion , then the total error of the sum should be on the order of @xmath127 . we can thus estimate @xmath128 . these results are discussed in section [ sub : newtonidentitiespractice ] . here we describe a systematic method for checking whether indeed all roots have been found that is based on the well known newton identities ( also known as newton - girard formulas ) that relate the coefficients of a polynomial ( which are up to sign the elementary symmetric functions of its roots ) to the power sums of all the roots . this method also makes it possible to locate missing roots in case not all were found ( see section [ sec : missingroots ] ) . the idea is to use these identities to compute from the coefficients of the polynomial @xmath3 the sums of the powers of all roots of @xmath3 , then to subtract from these the sums of powers of all roots found : the closer this difference is to @xmath125 , the better the accuracy of the roots found . of course it is difficult to find all coefficients of @xmath3 ( and many of them often have extremely large absolute values ) ; but in order to compute the first @xmath129 power sums of the roots , only the top @xmath129 coefficients are required ( beyond the leading coefficient that we usually scale to @xmath130 anyway ) . for large degrees , even the top few coefficients can become very large in practice , and this may present numerical challenges . we start by writing @xmath131 . we have the leading coefficient @xmath132 , and the subsequent coefficients @xmath133 are ( up to sign ) the elementary symmetric polynomials in the roots @xmath134 we need the power sums @xmath135 for @xmath136 . they can be computed from the coefficients @xmath133 by the newton identities as follows : @xmath137 and so on for subsequent coefficients . therefore , each @xmath138 can be computed recursively from the @xmath133 and the previously computed @xmath139 . in our applications , the polynomial @xmath3 is not given in coefficient form , so in the first step the coefficients @xmath133 have to be computed . only the coefficients @xmath140 are required . they can be found easily using the recursion that defines @xmath3 : if in the recursion in every step only the @xmath129 top - most coefficients are kept , then the final result will correctly yield the desired coefficients @xmath141 . to compute the coefficients @xmath133 , one needs @xmath71 recursion steps , each with omplexity @xmath142 ( or less if more efficient multiplication is implemented ) , for a total of @xmath143 . the theoretical power sums @xmath138 for @xmath136 are computed through the newton identities , which are a triangular system of linear equations of dimension @xmath129 , so they are evaluated in complexity @xmath144 . both steps are negligible compared to the computation of the actual power sums from the roots found , which has complexity @xmath145 . for the three families of polynomials , we computed the power sums for exponents @xmath146 , and compared them to the actual values computed using the newton identities . the differences are shown in figure [ fig : newtonidentityprecision ] . ( horizontal scale ) , for odd periods @xmath4 . top : centers of the mandelbrot set ; middle : periodic points of @xmath74 ; bottom : periodic points of @xmath75 . note that for the latter polynomials , the period @xmath147 graph includes three roots that were recovered using the newton identities . therefore , the values for @xmath148 are especially small ( these were used to reconstruct the three roots ) , while the others are relatively large ( due to the fact that the reconstructed roots have less precision than the others ) . the final @xmath89 graph is based on all roots found with maximal precision ( see section [ sub : newtonidentitiespractics ] ) . , scaledwidth=80.0% ] ( 0,0 ) ( -250,107)centers of the mandelbrot set ( horizontal scale ) , for odd periods @xmath4 . top : centers of the mandelbrot set ; middle : periodic points of @xmath74 ; bottom : periodic points of @xmath75 . note that for the latter polynomials , the period @xmath147 graph includes three roots that were recovered using the newton identities . therefore , the values for @xmath148 are especially small ( these were used to reconstruct the three roots ) , while the others are relatively large ( due to the fact that the reconstructed roots have less precision than the others ) . the final @xmath89 graph is based on all roots found with maximal precision ( see section [ sub : newtonidentitiespractics ] ) . , scaledwidth=80.0% ] ( 0,0 ) ( -250,94.5)periodic points of @xmath74 ( horizontal scale ) , for odd periods @xmath4 . top : centers of the mandelbrot set ; middle : periodic points of @xmath74 ; bottom : periodic points of @xmath75 . note that for the latter polynomials , the period @xmath147 graph includes three roots that were recovered using the newton identities . therefore , the values for @xmath148 are especially small ( these were used to reconstruct the three roots ) , while the others are relatively large ( due to the fact that the reconstructed roots have less precision than the others ) . the final @xmath89 graph is based on all roots found with maximal precision ( see section [ sub : newtonidentitiespractics ] ) . , scaledwidth=80.0% ] ( 0,0 ) ( -250,104)periodic points of @xmath75 the observed errors are between @xmath149 and @xmath150 , not surprisingly with increasing errors for higher degrees and higher powers . within reasonably expected numerical error ( discussed below ) , we see these computations as convincing confirmation that indeed all roots were found . one way to model the error in these tests is to assume that all roots found have numerical errors of average size @xmath126 , distributed equally and independently . specifically for the power sums with exponent @xmath151 , the sum is then a random walk of @xmath13 steps with average step size @xmath126 , so the accumulated error should be of size @xmath152 . figure [ fig : expectederror ] shows that the results for @xmath151 are consistent with the interpretation that typically @xmath153 . for @xmath154 , we encounter a precision issue for large numbers . let @xmath155 be the largest absolute values of the roots of a given polynomial ; we have @xmath156 for the mandelbrot centers , @xmath157 for periodic points of @xmath74 , and @xmath158 for periodic points of @xmath75 . then @xmath159 for some @xmath35 . these numbers increase exponentially with large @xmath65 and are eventually subtracted from each other , or from the exact coefficients that in our cases are ( gaussian ) integers , so ( for number formats with fixed lengths mantissae ) the absolute errors in the sums increase with @xmath160 ; this would lead to approximately linear graphs in the accuracies shown in figure [ fig : newtonidentityprecision ] , with slope @xmath161 . indeed , the average slopes are close to @xmath162 ( for the centers of the mandelbrot set ) , to @xmath163 ( for periodic points of @xmath74 ) , and to @xmath164 ( for periodic points of @xmath75 ) ; so in all three cases they are comparable to , and somewhat larger , than the predicted lower bounds of the accuracy . in any case , the most accurate tests for the achieved accuracy are those with low exponents @xmath65 . , for the three series of experiments ( vertical ) . the horizontal scale shows @xmath165 so that the degree is @xmath166 ( the periodic points have period @xmath165 , and the mandelbrot centers have period @xmath167 ) . observe that all values are of approximately constant size , mostly within a factor of @xmath168 . these data are consistent with the interpretation that all roots have been found with typical accuracy of at most @xmath169 . , scaledwidth=80.0% ] it might be interesting to report a few new challenges for these computations , coming from finite accuracy . one can use the newton identities in both directions : from the numerically computed power sums one can compute the coefficients and compare these with the true values from the polynomials , or do the computation backwards starting with the coefficients and compare the power sums . the latter approach , from coefficients to power sums , seems more natural in our case because we do most computations with ( gaussian ) integers , so without losing precisions . however , this involves _ much _ larger numbers ( our polynomials have very large coefficients , and we had to use integer arithmetic that can handle products of these without losing valid digits ) ; the computation in the other direction does not have the advantage of integers , but the numbers required are significantly smaller . another observation is that the very summation of the root powers loses valid digits . we had mentioned above that especially the high powers of the roots have limited absolute precision . in addition , when adding these large numbers , we often encounter very large sums ( consistent with the large predicted values from the newton identities ) . for instance , the powers @xmath170 for period @xmath89 of periodic points of @xmath74 sum up to about @xmath171 , which reduces the available absolute accuracy further . even when the power sum has small absolute value ( such as for @xmath172 for the same polynomial ) , the intermediate sums might become large before decreasing eventually ( especially when the roots found are sorted for instance by increasing real parts ) . we thus ran the addition a second time modulo @xmath130 and modulo @xmath35 , hence keeping track only of the fractional parts of the sums ( having checked that the integer part of the solution is consistent ) , so that all available accuracy can measure the difference to the exact integer value . the overall outcome of our three sets of experiments is that it is possible to find all roots of univariate polynomials of very large degrees in near - optimal complexity , both in terms of required newton iterations and in terms of computing time . all roots could be located for degrees up to 134 million , and the complexity per root is constant up to small logarithmic factors . the accuracy of all roots found is better than @xmath120 , comparable to the stopping condition @xmath173 . in one of the experiments , for periodic points of period @xmath89 of @xmath75 , three of the more than @xmath0 million roots were missed . these could be recovered with good precision by a method described in section [ sec : missingroots ] . a structural disadvantage of our iterated refinement newton root - finding method is that it does not come with an a priori guarantee that it will find all roots , and if in the end it turns out that some are missing , it is not clear where a refinement of starting points will recover the missing roots . this is different from the simpler linear scheme described in @xcite : there we usually start with @xmath15 equidistributed points on a single circle , and if some roots are missing , then this initial set of points can get refined for additional orbits . the crucial difference is that in the linear scheme from @xcite , every orbit starts at the same large circle surrounding all roots , so every orbit needs to run @xmath17 iterations before even getting close to the roots . in our iterated refinement newton algorithm , most orbits start very close to the roots , so far fewer iterations are required , but the refinement takes place away from the locus where we have good control . we gain a lot of speed at the expense of giving up guaranteed convergence . indeed , in some experiments it does happen that the iterated refinement newton algorithm finds almost all roots , but several of the more than one million roots are missing ( in one case that we describe in detail , three out of @xmath0 million roots were missing ) . roots may be missing either because the refinement was not sensitive enough , or because the roots were found but not declared sufficiently different from nearby roots that were reported . explicit deflation ( dividing the original polynomial by the product of all linear factors corresponding to the roots found ) is tempting but not an option : just building all coefficients from the roots found has quadratic complexity , so this involves much higher effort than all of root finding . our global scheme does not have a way to pinpoint where the missing roots were lost , and thus where additional starting points are required to spot these . here we describe three methods , presumably well known , that have low complexity and that find missing roots efficiently when the number of remaining roots is small : if the given polynomial @xmath174 has degree @xmath13 and @xmath129 roots are missing , then the complexity of this post - processing is essentially @xmath145 , up to logarithmic factors . we are grateful to dario bini , victor pan and michael stoll for helpful suggestions here , and to marvin randig for substantial computational contributions . the idea of implicit deflation is to evaluate newton s method for @xmath180 in terms of @xmath3 and the roots found : we have @xmath181 and thus @xmath182 ( for products of polynomials , the logarithmic derivative is additive ) . we can thus evaluate @xmath183 using only @xmath184 ( which is computed by the usual newton method for @xmath185 ) and the roots found so far . if @xmath186 , then the final sum dominates the complexity : every iteration step of @xmath187 has complexity essentially @xmath17 , while the number of necessary newton iterations depends only on the degree @xmath129 of @xmath179 . depending on the efficiency of root finding for this ( relatively ) low degree polynomial , the number of required newton iterations is between @xmath188 and @xmath144 , so the total complexity of finding the remaining roots is between @xmath189 and @xmath190 . a second method to locate the missing roots is the ehrlich - aberth method ( which is underlying the successful implementation ` mpsolve ` to find all roots ) . this method is a parallel iteration in @xmath13 variables , where the @xmath191-th component is the newton method of @xmath192 and the @xmath193 are the current approximations to all @xmath13 roots . as with implicit differentiation , one does not need to compute @xmath194 explicitly for the newton displacement , but uses an analog of . all @xmath13 coordinates are updated simultaneously . if @xmath195 roots are already known , then this method is an iteration in only @xmath129 variables . in particular , if @xmath196 , then implicit deflation equals the ehrlich - aberth method with @xmath197 coordinates fixed ; these two methods are thus closely related . a third method for locating missing roots is based on the newton identities described in section [ sub : newtonidentities ] that relate the coefficients of a polynomial ( which are up to sign the elementary symmetric functions of its roots ) to the power sums of all the roots . this method may be in less frequent use , even though it is based on very classical ideas . it allows one to construct a polynomial @xmath179 of degree @xmath129 the roots of which are exactly the missing roots of @xmath3 . this polynomial will be constructed in terms of its coefficients using sums of different powers of the roots as well as newton s identities . the idea is to use these identities to compute from the coefficients of the polynomial @xmath3 the sums of the powers of all roots of @xmath3 , then to subtract from these the sums of powers of all roots found : we thus obtain the sums of powers of the missing roots , and from these one can compute the coefficients of @xmath179 by applying the newton identities backwards ( where @xmath179 as above is the polynomial that has exactly the missing roots ) . we describe this method in somewhat more detail . now suppose the @xmath195 roots @xmath198 of @xmath3 have already been computed , and @xmath129 roots @xmath199 are missing . we can compute their power sums @xmath200 from the @xmath138 and the @xmath195 roots found . finally , we are looking for the polynomial @xmath201 where the @xmath202 are the coefficients of @xmath179 . the coefficients @xmath202 can be determined from the @xmath203 using the newton identities backwards : @xmath204 therefore , once we know @xmath205 , we can find @xmath179 in coefficient form and then apply any root finder to this polynomial of degree @xmath129 . in order to find the @xmath129 coefficients @xmath206 of @xmath179 , only the power sums @xmath207 are required in the newton identities backwards , and hence only the power sums @xmath208 and the coefficients @xmath209 are required . we described earlier in section [ sub : newtonidentities ] that these coefficients can be computed with little effort even when our polynomials @xmath3 are not given in coefficient form . the complexity of this method has been described in section [ sub : newtonidentities ] : it is dominated by the computation of the power sums @xmath210 , which requires @xmath145 operations . the remaining computations for @xmath129 missing roots can be performed in @xmath143 operations . one structural advantage of this method is that the original roots are required only for the simple computation of the power sums , not for the subsequent root finding iteration . in our applications , the roots were found on a different computer system than the one on which the missing roots were located , so not all @xmath195 previously found roots had to be transmitted , but only @xmath129 easily computed power sums ( it makes a substantial difference to transmit , or even to store , @xmath211 complex roots or @xmath212 power sums ) . this advantage comes with a certain disadvantage , though : the inaccuracies of all @xmath195 roots contribute equally to inaccuracies in the power sums , no matter how far they are from the missing roots ; in practice this limits the achievable accuracy of this method ( for the methods of implicit deflation or ehrlich aberth , the inaccuracy of roots that are far from missing roots becomes far less important ) . the two purposes of newton identity , to find missing roots and to check the roots found and estimate the accuracy achieved , can be combined : after @xmath129 missing roots have been reconstructed using the first @xmath129 coefficients , one can compute @xmath213 additional coefficients and hence @xmath213 additional power sums , and use these to verify that now all roots have been found with great accuracy . in this section ( performed in cooperation with marvin randig ) we briefly describe how the reconstruction of the 3 missing roots among the @xmath7 periodic points of period @xmath97 of @xmath96 works , and illustrate the results and some practical challenges . we are interested in the roots of @xmath214 for @xmath89 . the coefficients are gaussian integers , and the first few have the following values : @xmath215 with @xmath216 ; the subsequent @xmath133 with @xmath65 odd vanish , while those with @xmath65 even are huge . the respective power sums from the newton identities have comparable sizes . subtracting the sum of 134 million numerically computed roots to obtain power sums of the three missing roots is a certain challenge to the numerics involved . roots found ( separated into integer and fractional parts , and displaying less accuracy than used ) , and the last column shows the numerical accuracy achieved for the sum of all @xmath7 roots , including the 3 recovered roots and the predicted value from column 2 . the odd powers @xmath130 , @xmath212 , @xmath168 were used to reconstruct the three missing roots and thus show large computational accuracy . the remaining powers @xmath217 , @xmath218 , @xmath219 indicate that all roots have been found with about ten digits accuracy ( probably much better for most roots ) . , title="fig:",scaledwidth=95.0% ] nonetheless , in this case the three missing roots were found : two were close to @xmath35 , and one was close to @xmath125 . this was done initially using the power sums @xmath130 , @xmath72 , and @xmath212 . however , we have @xmath220 , so in the subtraction 9 valid decimal digits are lost , which is a severe limitation to accuracy . therefore , the computations were redone using the power sums @xmath130 , @xmath212 , and @xmath168 for which no serious cancellations occur ( the odd power sums of all roots should vanish ) . the results are shown in figure [ fig : threemissingroots ] . the third column shows that the odd power sums of the three missing roots must be equal to @xmath221 with an accuracy of @xmath222 , and indeed the three recovered missing roots are approximately @xmath223 the accuracy of these computations can be checked in the remaining power sums , as shown in the rightmost column in figure [ fig : threemissingroots ] . the odd powers @xmath130 , @xmath212 , @xmath168 show only that the computations were self - consistent : the three roots were recovered so that these power sums are correct within a computational accuracy of about 16 digits . -1 the other power sums provide independent tests . the fact that they all match to about ten valid decimal digits can be taken as experimental confirmation that all roots were found to at least this precision ( but there are problems when two missing roots are very close to each other ; see below ) . we believe that the initial @xmath211 roots were found with accuracy of at least @xmath120 as in all other computations , so that the power sums have accuracy of around @xmath224 , and the accuracy of the three recovered roots should be comparable to this latter value ( much lower than the remaining roots because of the accumulated error in the large sums ) . this is consistent with the power sum for @xmath225 , and we argued above why power sums for higher powers have less accuracy even when the roots themselves have small errors . ( for even powers , when the power sums were close to large gaussian integers , we again performed the computations twice , as described earlier : once to see that the sums are close to the desired gaussian integers , and once modulo gaussian integers to achieve higher absolute accuracy . ) for @xmath75 ( degree @xmath7 ) . left : a detail near the missing root @xmath226 . black shows points for which the newton method with @xmath48 failed to find a root . comparison with figure [ fig : newtonmarvin_1_19 ] ( same detail for the same polynomial , with @xmath227 ) shows that the problem was a sub - optimal stopping criterion : most black points actually converge to the missing root , which was however not recognized . right : the same polynomial with an even finer constant @xmath228 ( zoomed out ) : many more roots fail to be found . pictures by marvin randig.,title="fig:",scaledwidth=47.0% ] for @xmath75 ( degree @xmath7 ) . left : a detail near the missing root @xmath226 . black shows points for which the newton method with @xmath48 failed to find a root . comparison with figure [ fig : newtonmarvin_1_19 ] ( same detail for the same polynomial , with @xmath227 ) shows that the problem was a sub - optimal stopping criterion : most black points actually converge to the missing root , which was however not recognized . right : the same polynomial with an even finer constant @xmath228 ( zoomed out ) : many more roots fail to be found . pictures by marvin randig.,title="fig:",scaledwidth=47.0% ] with these approximations , the three missing roots were found also by a refinement of the original newton iteration ; compare figures [ fig : newtonmarvin_1_24 ] and [ fig : twomissingroots ] . improved approximations to these three roots are @xmath229 now with the same experimental accuracy as all other roots . a posteriori analysis reveals why the three roots were originally not found by our ad - hoc stopping criteria : for the first root , with our numerical accuracy we only achieve @xmath230 : we did have newton orbits that `` wanted to '' converge to the root , but the stopping criterion was too strict . the newton dynamics in figures [ fig : newtonmarvin_1_19 ] and [ fig : newtonmarvin_1_24 ] ( left ) illustrates this : both show the same detail of the newton dynamics around this missing root , and both were computed with the algorithm described except that in figure [ fig : newtonmarvin_1_19 ] the requirement that @xmath231 was relaxed to @xmath232 . the black points in figure [ fig : newtonmarvin_1_19 ] are points that do not converge to roots with the initial choice of @xmath233 . estimates using ( * ? ? ? * lemma 4 ) prove that the improved approximation to the first root is indeed close to a root with error at most @xmath234 ( probably better ) . the other two roots ( both near @xmath35 ) were missed because they were closer than @xmath52 from two other roots : in the first case , there was another root found at distance @xmath235 , and in the second case at distance @xmath236 , both somewhat smaller than @xmath237 in absolute value . . the two missing roots near @xmath35 are indicated by blue dots together with the two nearby roots that were found ( manually inserted , not to scale : the distance between the pairs is @xmath238 , while the distance within both pairs is close to @xmath237 ) . picture by marvin randig.,width=257 ] as we pointed out , the values of the constants in our algorithm were chosen in an ad - hoc fashion and obviously have to be adjusted for polynomials of extremely large degrees but the underlying newton method is stable enough that it works well even with these ad - hoc criteria . once the three missing roots were found to full precision as in , we compared their values with the approximations from the newton identities as in . for the root near @xmath107 , the error was @xmath239 , almost on the nose equal to the estimated value . however , for the two roots near @xmath35 , the error turned out to be equal to @xmath240 for both , much larger than expected . we provide an explanation below . observe first that the fact that both roots have errors with almost equal absolute values is due to the fact that they must match the sums of the first powers , so the deviations must be opposite to each other within the precision of the power sums . the fact that the last two missing roots are very close to each other has to do with a particular property of our polynomial : all the roots we are looking for are periodic points of @xmath96 , so as a set they are invariant by @xmath241 . they are equidistributed with respect to harmonic measure on the julia set of @xmath241 , but all roots near @xmath107 are contracted by the quadratic map @xmath241 that has a critical point at @xmath107 , so the roots must be very close to each other near @xmath242 . the pairs of nearest roots are thus to be expected to be close to @xmath108 , and the trouble was caused by two such pairs . after the missing root near @xmath107 is recovered numerically with expected precision , the two missing roots near @xmath108 form a quadratic equation with an ( almost ) double root , given with some precision @xmath243 . in such a case , the roots only have precision @xmath244 , and this is what we experience here . it is an interesting fact that in the reconstruction of the last two missing roots , an extremely ill - conditioned quadratic equation causes substantial errors , while newton s method itself is able to find all @xmath7 roots ( many of them much closer to each other than this pair ) , even the three initially missing roots and the nearby roots close to them , without any issues of ill - conditioning ( except that we had to adjust our ad - hoc threshold parameters ) . of course , ill - conditioning is an ubiquitous experience , but it is remarkable how well newton can handle it . bla bollobs , malte lackmann , and dierk schleicher , _ a small probabilistic universal set of starting points for finding roots of complex polynomials by newton s method_. mathematics of computation * 82 * ( 2013 ) , 443457 . dierk schleicher and robin stoll , _ newton s method in practice : finding all roots of polynomials of degree one million efficiently_. journal of theoretical computer science , to appear ( 2016 ) .	we present a practical implementation based on newton s method to find all roots of several families of complex polynomials of degrees up to @xmath0 million so that the observed complexity to find all roots is between @xmath1 and @xmath2 ( measuring complexity in terms of number of newton iterations or computing time ) . all computations were performed successfully on standard personal computers made in 2010/2011 , using only a single processor core .
the degree of understanding of quantum correlations ( qc ) underlies our current picture of nature @xcite . it has been recently found that there exist important manifestations of the quantumness of correlations in composite systems that are different from those of entanglement - origin ( eo ) and that may be relevant in quantum information technologies @xcite . the quantifier of these non - eo correlations is called the quantum discord ( qd ) @xmath1 and is based , for a bipartite system , on shannon s mutual information . we are thus speaking of an information - theoretic ( it ) tool . for pure states qd does not add any qcs , but that is not the case for mixed states . the @xmath6concept , advanced in the pioneer paper by ollivier and zurek @xcite , quantifies : i ) the minimum change in the state of the system and ii ) the information on one of its parts induced by a measurement of the other one . if the state has only classical correlations , @xmath1 vanishes , which implies that the quantum discord concept somehow quantifies the `` correlational - quantumness '' . it has been evaluated for several families of states both in its original form and in variously altered versions and generalizations . a particularly compelling instance expresses the qd - notion in terms of conditional density operators @xcite . interesting operational qd - interpretations have also been advanced @xcite . evaluating qd requires a rather involved optimization procedure , analytical expressions being known in just a few instances @xcite . @xmath1 is built up as an entropic difference , the difference between a quantum entropic measure and its classical counterpart , which is derived from local measurements on one or both of the participant subsystems . its amount is a new feature @xmath7 of the quantum state @xmath8 , which in turn induces a @xmath6amount `` ordering '' for states of the form @xmath9 , for instance . now , it - tools come in many distinct varieties . given the immense body of literature that has been generated in the past two decades concerning physically motivated statistical formalisms based on generalizations of shannon s information measure ( see @xcite and references therein ) , it seems both natural and necessary to tackle the qd issues from this generalized angle , in the hope of gaining interesting insights , and , in particular , so as to establish the invariance or not of the discord - induced order under a change of the prevailing statistics , from shannon s to its many rivals . to show that this is indeed a fruitful endeavor is the aim of the present paper , in which a generalization of the quantum discord _ concept _ , in the context of generalized statistics , will be advanced and the `` ordering - question '' answered . in section 2 we introduce our conceptual qd - extension , discussing its main properties in sect . next we present some results for general bipartite states , focusing attention on analytical results . we also perform numerical simulations for random bipartite states ( sect . finally , some conclusions are drawn in section 5 . tsallis power - law @xmath11-entropy was introduced in @xcite as an extension of the shannon entropy as follows @xcite @xmath12 where the @xmath11-logarithm is defined by @xmath13 , @xmath14 is the probability distribution of the pertinent random variable @xmath15 , and @xmath11 is any nonnegative real number . tsallis s entropy converges to shannon s in the limit @xmath16 . @xmath17 plays a fundamental role in recent developments of statistical mechanics @xcite . the generalization has indeed received a lot of attention in the last years , with about 2000 papers containing interesting results and useful applications , many of them in the complex systems area @xcite but also in connection with a variegated family of quantum mechanical settings ( see , for example @xcite ) . in what follows we retrace the developments of ref . @xcite in a tsallis , @xmath18context . thus , just by setting @xmath19 we recover the ollivier - zurek quantities . we begin then with the mutual information , defined as @xmath20 and the following chain rule holds @xcite : @xmath21 where the conditional entropy reads @xmath22 the chain rule gives the relation between a conditional entropy and a joint entropy . using this relation we can define another , classically equivalent , expression for the mutual information @xmath23 the @xmath24 difference is of the essence for ollivier - zureck goals , after expressing the two quantifiers in quantal fashion . let us then do the same with @xmath25 and @xmath26 . @xmath25 can be easily generalized defining appropriate density matrices for the quantum systems , @xmath27 , @xmath28 , and @xmath29 , and applying then the @xmath11-generalization of von neumann s entropy @xmath30 . one has @xmath31 to generalize the @xmath32expression , following @xcite , we focus on a perfect measurements of @xmath28 defined by a set of projectors @xmath33 such that @xmath34 . accordingly , @xmath35 where @xmath36 with the state of @xmath8 given , once the measurement is performed , by @xmath37 and @xmath38 . the two classical expressions for the standard mutual information we have presented above are identical , but this is not necessarily so in the quantum case actually , the quantum discord is defined as the _ minimum _ possible difference between the two possibilities , given by an _ optimum _ set of @xmath33 @xcite . thus , we are to be concerned here with what happens to the expressions @xmath25 and @xmath26 . introduce the quantity @xmath39 we define now our quantum q - discord as the difference @xmath40 this quantum q - discord is the minimum of the difference between eq . ( [ imutua ] ) and eq . ( [ jmutua ] ) . we normalize this measure via a trivial re - scaling in order to compare , in an adequate way , different quantities : @xmath41 for similar reasons , @xmath42 denotes the logarithm of base 2 . in what follows @xmath43 . note that @xmath44 . we see that @xmath45 for @xmath46 , and this might be related to the concavity of the @xmath11-conditional entropy @xmath47 with respect to @xmath29 . indeed , @xmath48 becomes negative if @xmath11 grows from @xmath49 , negativity increasing with @xmath11 . quite convenient is the particular case @xmath50 , since it requires only to compute the traces of @xmath51 , and matrix - diagonalization is avoided , making computations more efficient . taking the limit @xmath52 of the normalized measure we obtain @xmath53 to study the positivity of the q - discord we consider separately two cases : pure states and mixed states . for pure states , that is @xmath54 , the q - discord takes the form @xmath55 and the quantum q - discord coincides with the reduced ( quantum ) tsallis entropy . we can easily verify this fact by casting @xmath56 in its schmidt decomposition form @xmath57 . thus , since the q - entropy is positive for all @xmath11 , @xmath58 in the case of mixed states our q - discord is positive only for values of @xmath11 in @xmath59 . in order to demonstrate the positivity of the q - discord for mixed arbitrary states we follow @xcite and consider the proposition : @xmath60 with @xmath61 . now , @xmath62 is block diagonal and as in the @xmath19 case , doing a block by block analysis the proposition can be proved . now we need to verify the inequality : @xmath63 by recourse of the previous preposition we can establish the following relation for any measurement @xmath33 @xmath64 and by the ( conjectured ) concavity ( see below ) of the conditional entropy ( @xmath65 ) with respect to @xmath66 for @xmath46 we are led to @xmath67 the set of states in an @xmath69-dimensional hilbert space can be regarded as a product - space of the form @xcite , @xmath70 where @xmath71 stands for the family of all complete sets of ortho - normal projectors @xmath72 ( @xmath73 the identity matrix ) , and @xmath74 is the convex set of all real @xmath68tuples of the form @xmath75 any state in @xmath76 takes the form @xmath77 in order to explore @xmath78 we introduce an appropriate measure @xmath79 on this space . such a measure is required to compute volumes within @xmath78 , as well as to determine what is to be understood by a uniform distribution of states on @xmath80 . an arbitrary state @xmath66 of our @xmath69-dimensional hilbert space can always be expressed as a product of the form @xmath81 u^{\dagger}.\ ] ] here @xmath82 is an @xmath83 unitary matrix and @xmath84 $ ] is an @xmath83 diagonal matrix whose diagonal elements are , precisely , our above defined @xmath85 . the group of unitary matrices @xmath86 is endowed with a unique , uniform measure , known as the haar s measure , @xmath87 @xcite . on the other hand , the @xmath69-simplex @xmath74 , consisting of all the real @xmath69-uples @xmath88 appearing in ( [ udot ] ) , is a subset of a @xmath89-dimensional hyperplane of @xmath90 . consequently , the standard normalized lebesgue measure @xmath91 on @xmath92 provides a measure for @xmath74 . the aforementioned measures on @xmath86 and @xmath74 lead to a measure @xmath79 on the set @xmath93 of all the states of our quantum system @xcite , @xmath94 if one needs to randomly generate mixed states , this is to be done according to the measure ( [ memu ] ) . here we attempt a numerical verification of the concavity of the conditional entropy for @xmath46 , that is , @xmath96+(1-t)[s_q(\xi)-s_q(\xi_a)]\}\geq 0,\ ] ] where @xmath97 , @xmath27 is the reduced density matrix corresponding to @xmath66 , @xmath98 ( @xmath99 ) the reduced density matrices of @xmath100 ( @xmath101 ) and , finally , @xmath102 . for different values of @xmath103 for which the concavity is verified . inset : @xmath104 for different @xmath105 . the curves are left- shifted . all curves were constructed using of the order of @xmath106 ( numerically generated ) states . all depicted quantities are dimensionless.[pdfs ] ] the concavity of the standard conditional entropy ( @xmath19 ) was proved in @xcite by assuming the validity of a lemma by lieb @xcite . the proof is rather difficult even in this case @xcite . as a first step we evaluate numerically the inequality ( [ deltaq ] ) by generating random states in an @xmath69-dimensional hilbert space . in order to assess , for these randomly generated states , how the concavity - requirement is satisfied , we evaluate ( [ deltaq ] ) for a large enough number of simulated states ( @xmath100 and @xmath101 ) . we set @xmath107 for the dimension of the state - space in all simulations and we randomly generate @xmath108 $ ] . we investigate the positivity of @xmath109 , upon which the concavity of the conditional q - entropy is based , by constructing the probability distributions for the values of @xmath110 . the corresponding distributions , for different values of @xmath11 , are depicted in fig . [ pdfs ] . in the inset we plot the probability distribution of @xmath109 for @xmath50 and @xmath111 . the curves are constructed using of the order of @xmath106 states . these simulations provide us with strong evidence about the validity of the conjecture advanced above on the concavity of the conditional q - entropy and , consequently , on the positivity of the quantum q - discord for @xmath112 . let us now investigate the relation between the q - discord and its original counterpart for different sorts of states . we begin with bell diagonal states . these are two - qubit states with maximally - mixed reduced - density matrices and have the form @xmath113 where @xmath114 are real constants constrained by certain conditions ( in order to have a well defined density operator @xmath66 ) and @xmath115 s are the pauli operators . let @xmath116 $ ] , @xmath117 be the eigenvalues of @xmath66 @xmath118 the marginal states of @xmath66 are @xmath119 and @xmath120 . thus , the quantum q - mutual information of @xmath66 is @xmath121 and @xmath122,\ ] ] where @xmath123 . we find , for a general ( bell - diagonal ) two - qubit state , @xmath124+\sum_i\lambda_i^q\ln_q\lambda_i.\ ] ] let us specialize ( [ ana1 ] ) to the particular instance @xmath125 , i.e. , the celebrated werner states , @xmath126,\ ] ] with @xmath127 . by following @xcite one easily obtains @xmath128\nonumber\\&+&3\left(\frac{1-c}{4}\right)^q\ln_q\frac{1-c}{4}+\left(\frac{1 + 3c}{4}\right)^q\ln_q\frac{1 + 3c}{4},\end{aligned}\ ] ] and , as seen in fig . [ wernerstates ] , positivity prevails for the prototype - mixed state . in fig . [ wernerstates ] we plot the ( normalized ) q - discord as a function of the state - parameter @xmath129 for different values of @xmath11 and also as a function of the mixedness - degree as given by the linear entropy @xmath130,\ ] ] trivially related to the purity @xmath131 of a state via @xmath132 as expected , an inverse relationship between mixedness - degree and quantum correlations is displayed . we remark on the single - valuedness of the werner - relation between q - discord and mixedness , even for @xmath19 for werner state , as a function of @xmath129 , for different values of @xmath11 . ( b ) @xmath2 for werner states , as a function of the degree of mixedness measured by the linear entropy , for different values of @xmath11 . all depicted quantities are dimensionless.[wernerstates],title="fig : " ] for werner state , as a function of @xmath129 , for different values of @xmath11 . ( b ) @xmath2 for werner states , as a function of the degree of mixedness measured by the linear entropy , for different values of @xmath11 . all depicted quantities are dimensionless.[wernerstates],title="fig : " ] we also will study the quantum q - discord for the following one - parameter states for @xmath133 states , as a function of @xmath133 , for different values of @xmath11 . ( b ) @xmath2 for @xmath133 states , as a function of the degree of mixedness , for different values of @xmath11 . all depicted quantities are dimensionless.[alphastates],title="fig : " ] for @xmath133 states , as a function of @xmath133 , for different values of @xmath11 . ( b ) @xmath2 for @xmath133 states , as a function of the degree of mixedness , for different values of @xmath11 . all depicted quantities are dimensionless.[alphastates],title="fig : " ] @xmath134 where @xmath135 . let @xmath136 . the q - discord becomes @xmath137 + 2\left(\frac{1-\alpha}{2}\right)^q\ln_q\frac{1-\alpha}{2}+\alpha^q\ln_q\alpha.\ ] ] in fig . [ alphastates ] we depict the quantum q - discord as a function of the state s parameters for different values of @xmath11 and also plot it as a function of the linear entropy . positivity again prevails . the single - valuedness between discord and mixedness is lost for these states . of two @xmath133 states as a function of the parameter @xmath11 . a non trivial ordering relation is found . all depicted quantities are dimensionless.[alphasigno ] ] in fig . [ alphasigno ] we display the difference between the q - discord of two @xmath133-states ( corresponding to @xmath139 and @xmath140 , respectively ) , as a function of @xmath11 . this difference takes negative or positive values depending on the range of @xmath11 . this is indeed a novel feature . a relation of order for quantum states based on the discord - concept can not univocally be established , because it depends on the entropic quantifier one chooses to employ . in other words , the quantal correlations that the discord quantifies are seen in different manners by distinct entropic quantifiers . _ this lack of uniqueness is the leit - motif of the present considerations_. as a last particular kind of special state to be analyzed , consider the two - parameters state @xmath142 where @xmath135 and @xmath143 . we display the @xmath144discord versus the @xmath145discord for this state in fig . [ abstate ] . a strong correlation is exhibited between the two @xmath18measures . this could be taken as evidence that changing @xmath11 from its original @xmath145value does not per se modify the overall manner in which q - discord quantifies quantum correlations . ( q=2 ) vs @xmath1 for @xmath146-like states . all depicted quantities are dimensionless.[abstate ] ] here , we focus our discussion on general ( pure or mixed ) states of two qubits . for such system we can parametrize the basis of the measurement by @xmath147 and @xmath148 , latexmath:[\[\begin{aligned } \|\psi\rangle&=&\cos(\theta)\|0\rangle+e^{i\phi}\sin(\theta)\|1\rangle\\\nonumber as a function of @xmath1 for randomly generated two - qubit states ( a ) @xmath150 , ( b ) @xmath50 . all depicted quantities are dimensionless.[qdranstates],title="fig : " ] as a function of @xmath1 for randomly generated two - qubit states ( a ) @xmath150 , ( b ) @xmath50 . all depicted quantities are dimensionless.[qdranstates],title="fig : " ] we numerically search the @xmath147 - @xmath148 space for the set of values that maximize eq . ( [ optimization ] ) . the resultant density operator @xmath151 , when such measurements are performed on subsystem @xmath152 , is @xmath153 where each complete set , composed of two elements , of possible measurements is defined as follows , @xmath154 . the curves were constructed using @xmath155 generated random states . all depicted quantities are dimensionless . [ qdpdf ] ] we randomly generate states uniformly distributed according to the measure @xmath156 and by recourse of the previously described optimization procedure . of course , we numerically search for @xmath147 and @xmath148 and compute the q - discord for these states . in fig . [ qdranstates ] we display the correlation between the discord and the q - discord for different values of @xmath11 . negative values of the q - discord are depicted in the plot for the case @xmath105 ( @xmath50 ) . figure [ qdpdf ] depicts the probability distribution of finding a given value of q - discord in the whole space of two - qubit states for @xmath157 . we also compute the difference between the q - discord and the discord between pairs of randomly generated states @xmath66 and @xmath100 . in fig . [ qddif ] we plot the resultant differences of the q - discord versus similar differences ( for the same pair of states ) corresponding to 1-discord . this plot depicts the pertinent results . overall , our numerical simulations confirm the conclusions reached by the analysis of special kinds of states . as a function of @xmath158 ( a ) @xmath150 , ( b ) @xmath50 . all depicted quantities are dimensionless . [ qddif],title="fig : " ] as a function of @xmath158 ( a ) @xmath150 , ( b ) @xmath50 . all depicted quantities are dimensionless . [ qddif],title="fig : " ] we have introduced a new family of quantum discord measures that quantifies quantum correlations based in the chain rule relating the i ) conditional- and ii ) joint - tsallis entropies . via two types of study 1 . there is a strong correlation between the `` new '' q - discord and the original one of ollivier and zureck . 2 . however , an order - relation for quantum states based on discord lacks unicity because it definitely depends on the quantifier one chooses to employ . this means that @xmath11-discord functionals corresponding to different values of @xmath11 measure different aspects of the non - classicality ( quantumness ) of correlations . this work was partially supported by the projects fqm-2445 and fqm-4643 of the junta de andalucia and the grant fis2011 - 24540 of the ministerio de innovacin y ciencia ( spain ) . a.p.m . acknowledges support by genil through ytr - genil program . acknowledges support from the senior grant cei bio - tic genil - spr . l. amico , l. fazio , a. osterloh , v. vedral , rev . mod . phys . * 80 * ( 2008 ) 517 . i. bengtsson , k. zyczkowski , geometry of quantum states : an introduction to quantum entanglement , cambridge univ . press , cambridge , 2006 . v. vedral , phys . lett * 90 * ( 2003 ) 050401 . j. maziero , l.c . celri , r.m . serra , v. vedral , phys . a * 80 * ( 2009 ) 044102 . j. oppenheim , m. horodecki , p. horodecki , r. horodecki , phys . lett * 89 * ( 2002 ) 180402 . m. ali , a.r.p . rau , g. alber , phys . a * 81 * ( 2010 ) 042105 . s. furuichi , j. math . phys . * 47 * ( 2006 ) 023302 . a. wehrl , rev . * 50 * ( 1978 ) 221 . e. lieb , adv . math . * 11 * ( 1973 ) 267 . k. zyczkowski , p. horodecki , a. sanpera and m. lewenstein , phys . a * 58 * ( 1998 ) 883 .	the notion of quantum discord introduced by ollivier and zurek [ phys . rev . lett * 88 * , 017901 ( 2001 ) ] ( see also henderson and vedral [ j. phys . a * 34 * , 6899 ( 2001 ) ] ) has attracted increasing attention , in recent years , as an entropic quantifier of non - classical features pertaining to the correlations exhibited by bipartite quantum systems . here we generalize the notion so as to encompass power - law q - entropies ( that reduce to the standard shannon entropy in the limit @xmath0 ) and study the concomitant consequences . the ensuing , new discord - like measures we advance describe aspects of non - classicality that are different from those associated with the standard quantum discord . a particular manifestation of this difference concerns a feature related to order . let @xmath1 stand for the standard , shannon - based discord measure and @xmath2 for the @xmath3 one . if two quantum states @xmath4 are such that @xmath5 , this order - relation does not remain invariant under a change from @xmath1 to @xmath2 .
in the standard model of elementary particles the fermion and gauge boson masses are generated due to the interaction of these particles with elementary higgs scalar bosons . despite its success there are some points in the model as , for instance , the enormous range of masses between the lightest and heaviest fermions and other peculiarities that could be better explained at a deeper level . the nature of the higgs boson is one of the most important problems in particle physics , and there are many questions that may be answered in the near future by the lhc experiments , such as : is the higgs boson , if it exists at all , elementary or composite ? what are the symmetries behind the higgs mechanism ? among the priorities of the higgs boson search at the lhc experiments , is the measurement of its mass , width , spin and cp eigenvalues . the measurement of the higgs boson couplings and , particularly , its self - couplings will also be quite important , once they may unravel all the subtleties of the mechanism of electroweak symmetry breaking @xcite . it may be possible to measure the higgs boson self - coupling at the lhc in the case of a light higgs boson@xcite , but this will barely be possible if the boson mass is larger than @xmath1 gev . there are many variants for the higgs mechanism . our interest in this work will be focused in the models of electroweak symmetry breaking via strongly interacting theories of technicolor type @xcite . in these theories the higgs boson is a composite of the so called technifermions , and at some extent any model where the higgs boson is not an elementary field follows more or less the same ideas of the technicolor models . in extensions of the standard model the scalar self - couplings can be enhanced , like in the supersymmetric version@xcite . if the same happens in models of dynamical symmetry breaking , as far as we know , has not been investigated up to now . the beautiful characteristics of technicolor ( tc ) as well as its problems were clearly listed recently by lane@xcite . most of the technicolor problems may be related to the dynamics of the theory as described in ref.@xcite . although technicolor is a non - abelian gauge theory it is not necessarily similar to qcd , and if we can not even say that qcd is fully understood up to now , it is perfectly reasonable to realize the enormous work that is needed to abstract from the fermionic spectrum the underlying technicolor dynamics . the many attempts to build a realistic model of dynamically generated fermion masses are reviewed in ref.@xcite . most of the work in this area try to find the tc dynamics dealing with the particle content of the theory in order to obtain a technifermion self - energy that does not lead to phenomenological problems as in the scheme known as walking technicolor@xcite . the idea of this scheme is quite simple . first , remember that the expression for the tc self - energy is proportional to @xmath2 , where @xmath3 is the tc condensate and @xmath4 its anomalous dimension . secondly , depending on the behavior of the anomalous dimension we obtain different behaviors for @xmath5 . a large anomalous dimension may solve the problems in tc models . in principle we could deal with many different models , varying fermion representations and particle content , finding different expressions for @xmath5 and testing them phenomenologically , i.e. obtaining the fermion mass spectra without any conflict with experiment . usually the walking behavior is obtained only with a large number of technifermions , although there are recent proposals where the walking behavior is obtained for a very small number of fields with the introduction of technifermions in higher dimensional representations of the technicolor gauge group@xcite . as the dynamics in models of dynamical symmetry breaking can be so different from qcd , it is interesting to investigate the behavior of the dynamical higgs boson self - coupling . it could be possible that in these models the self - coupling is enhanced and easy to measure than the standard one , even considering that the dynamical higgs boson will be heavier than @xmath1 gev . in this work we will consider a very general ansatz for the technifermion self - energy that was introduced in the ref.@xcite . this ansatz interpolates between all known forms of technifermionic self - energy . as we vary some parameters in our ansatz for the technifermionic self - energy we go from the standard operator product expansion ( ope ) behavior of the self - energy to the one predicted by the extreme limit of a walking technicolor dynamics , i. e. @xmath6 @xcite.we will discuss the general properties of the trilinear higgs coupling based on this ansatz . in principle the trilinear higgs self - coupling can be measured directly in the higgs boson pair production at the lhc via the gluon - gluon fusion mechanism . in this way we can predict the general behavior of this cross section in technicolor models . as , in non - abelian gauge theories , the fermion self - energy is related to the bethe - salpeter wave - function of the composite scalar boson @xcite , we believe that our result could also be extended to any composite higgs model . this paper is organized as follows : in sec . ii we compute the trilinear self - coupling of a composite higgs boson assuming the anzatz for the fermionic self - energy shown in ref.@xcite . in the sec . iii we review the self - couplings of the standard model fundamental higgs field and compare them with the results shown in the previous section . in the sec . iv we compute the cross section for the @xmath7 reaction in the case of the standard model and for a composite higgs boson . finally in the sec.v we draw our conclusions . using ward identities we can show the couplings of the scalar boson to fermions to be@xcite g^_a _ ( p+q , p ) = - [ fsc ] where @xmath8 , @xmath9 is a @xmath10 matrix , and @xmath11 is a matrix of fermionic self - energies in weak - isodoublet space . as in ref.@xcite we assume that there is a scalar composite higgs boson that couples to the fermionic self - energy which is saturated by the top quark@xcite . specifically , we assume that the scalar - to - fermion coupling matrix at large momenta is given by @xmath12 , where we do not attempt to distinguish between the two fermion momenta @xmath13 and @xmath14 , since , in all situations with which we will be concerned , @xmath15 . therefore the coupling between a composite higgs boson with fermions at large momenta is given by _ _ hff(p)g(p , p ) ~-(p^2 ) [ fh ] where @xmath16 is the fermionic self - energy . the trilinear higgs boson coupling in technicolor models will be dominated by loops of heavy fermions that couple to the scalar higgs particle as predicted by eq.([fh ] ) @xcite . our purpose in this section is to obtain an expression for the trilinear higgs boson coupling using the ansatz _ a(p^2 ) ~__tc ( ) ^^- , [ sig ] which was proposed in the ref.@xcite . this choice interpolates between the standard ope result for the technifermion self - energy , which is obtained when @xmath17 , and the extreme walking technicolor solution obtained when @xmath18 @xcite , i.e. this is the case where the symmetry breaking is dominated by higher order interactions that are relevant at or above the tc scale , leading naturally to a very hard dynamics @xcite . as we have pointed out in ref . @xcite only such kind of solution is naturally capable of generating a large mass to the third fermionic generation , which has a mass limit almost saturated by the top quark mass . moreover , as also claimed in the second paper of ref.@xcite , there are other possible reasons to have @xmath19 , as the existence of an infrared fixed point and a gluon ( or technigluon ) mass scale @xcite , which , actually , are related possibilities @xcite . the yukawa top quark coupling to the higgs boson is large , no matter we are considering the composite or the fundamental standard model higgs boson , and is the one that dominates the process that we will be considering in the next sections . the main difference is that in the composite case the trilinear coupling is a function of this yukawa coupling as remarked in ref.@xcite . as considered in ref.@xcite and many others dealing with dynamical symmetry breaking models it is usually assumed that such calculations are not spoiled by higher order corrections . it is interesting that many technicolor models make use of the existence of a non - trivial fixed point ( or a quasi - conformal theory ) to cure their phenomenological problems@xcite , and exactly for this possibility brodsky has been claiming that it will be possible to built an skeleton expansion that could allow to capture the non - perturbative effects in a reliable way@xcite . in the eq.([sig ] ) the scale , @xmath20 is related to the technicolor condensate by @xmath21 . we defined @xmath22 , @xmath23 with @xmath24 , and @xmath25 is the quadratic casimir operator given by @xmath26\ ] ] where @xmath27 , are the casimir operators for technifermions in the representations @xmath28 and @xmath29 that condensate in the representation @xmath30 , @xmath31 is the coefficient of the @xmath32 term in the technicolor @xmath33 function . we can determine one expression for the trilinear coupling for any theory where the higgs boson is composite by considering the diagram shown in fig.(1 ) . the contribution of fig.([fig1 ] ) is certainly the dominant one @xcite . assuming the coupling of the scalar boson to the fermions given by the eq.(1 ) , and with the fermion propagator written as s_f(p ) = we find that ^t__hhh = ( ) ^_0 . [ tri ] where @xmath34 is the number of technifermions included in the model . considering the ansatz given by the eq.([sig ] ) , and introducing it into eq.([tri ] ) , we obtain ^t__hhh ( ) ^4__tc(__tc)^4i(p^2 ) [ eq5 ] with @xmath35 to compute this last expression we have used the following mellin transform [ 1 + a]^-= ^_0dzz^-1e^-z(b)^-az . after performing the @xmath36 integration in eq.([eq5 ] ) , we can write this equation as ^t__hhh ( ) ^_0 . [ eq7 ] we will present our analysis of @xmath37 for two different regions of the parameter @xmath38 . we will start with the case @xmath39 . therefore we can make the following expansion in eq.([eq7 ] ) . than eq.([eq7 ] ) can be cast in the form ^t0__hhh ( ) & & retaining only the first two terms in the @xmath38 expansion and performing the @xmath40 integration , we finally can write ^t0__hhh ( ) . [ eq9 ] when @xmath41 , we can consider a similar expansion , and following the same steps we obtain ^t1__hhh ( ) [ eq10 ] the above expressions for the trilinear higgs coupling are quite dependent on the scale @xmath20 . this is not the best formula to compute this coupling , since @xmath20 , which in principle is related to the value of the dynamical technifermion mass at the origin , is not directly fixed by the symmetry breaking of the standard model . a more appropriate quantity that can be used to describe this coupling is the technipion decay constant , which is fixed by the @xmath42 and @xmath43 gauge boson masses . considering our comments in the previous paragraph we will express the trilinear higgs coupling as a function of the technipion decay constant ( @xmath44 ) instead of the scale @xmath20 . @xmath44 can be computed through the known pagels and stokar relation@xcite f^2 _ = ^_0 where @xmath45 is the technicolor number . we compute the technipion decay constant using the ansatz eq.([sig ] ) . after some calculation we obtain the following expression for @xmath44 f^2 _ _ = ^2__tcf(k ) [ eq12 ] where f(k ) = csc with @xmath46 to obtain this expression we have assumed the scaling law @xmath47@xcite . to be consistent with eqs.([eq9 ] ) and ( [ eq10 ] ) , we also need to expand eq.([eq12 ] ) for @xmath39 and @xmath41 . in this case , we obtain f^2 _ _ = ^2__tc with @xmath48 finally , assuming this last equation , we can write the eqs.([eq9 ] ) and ( [ eq10 ] ) in the form ^t__hhh= 3n_ff ( ) [ eq15 ] where for convenience we defined @xmath49}{(1 - 5\alpha)^2}\,\,\,\,{\rm when}\,\,\,\ , \alpha\approx 0 \\ \\ \frac{3}{4}\frac{[1 - 4(\alpha - 1)/a ] } { ( 1 - \alpha/2)^2 } \,\,\,\,{\rm when}\,\,\,\ , \alpha\approx 1 \end{array}\right.\ ] ] and will assume @xmath50 doublets of technifermions @xmath51 , and in this work we will be assuming @xmath52 . ] our ansatz for the fermionic self - energy is a very general one . no matter which is the theory ( technicolor or any of its variations ) the self - energy will be limited to the expressions obtained from eq.([sig ] ) for @xmath38 in the range @xmath53 $ ] , even the scenario proposed in ref.@xcite will be described by such expression . in the next section we will compare these expressions for the trilinear composite higgs boson self - coupling with the one of the standard model fundamental higgs boson . in this section we review the expression for the trilinear coupling in the case of the standard model fundamental higgs boson , and compare it to the ones found in the previous section . we start writing the expression of the higgs boson potential in the standard model v ( ) = -^2^+ ( ^)^2 . the self - couplings are uniquely determined in the standard model by the mass of the higgs boson , which is related to the quadrilinear coupling @xmath55 by the following expression @xmath56 after introducing the physical higgs field h in the neutral component of the doublet @xmath57 we can write the potential as v(h ) = h^2 + h^3 + h^4 . the multiple higgs couplings can be derived from the potential @xmath58 , and the trilinear and quadrilinear couplings of the higgs field h are given by & & _ _ 3h= 3_0 + & & _ _ 4h= 3 ^ 2_0 . [ e18 ] to obtain these expressions we assumed the normalization employed in ref.@xcite , where @xmath59 . in the case of a composite higgs boson it is possible to show that its mass can be expected to be of the following order @xcite : @xmath60 this result is independent of the dynamics and is originated from the similarity between the schwinger - dyson equation for the technifermion self - energy and the bethe - salpeter equation for the scalar channel @xcite . of course , as discussed in the previous section , we will write @xmath61 as a function of @xmath62 instead of @xmath20 . to compare the results of the previous section with the couplings shown above we can write the couplings for the composite higgs boson as a function of its mass . assuming the mass relation given above , considering the eq.([eq15 ] ) and rewriting it in terms of the parameter @xmath63 , we obtain in the case of @xmath64 ^t0__3h = ( ) , [ e19 ] and in the case when @xmath65 we obtain ^t1__3h = ( ) _ 0 . [ e20 ] where @xmath66 . in the fig.(2 ) the behavior of the trilinear higgs couplings is plotted as a function of the higgs boson mass . the solid line represents the contribution of the fundamental higgs boson , i. e. the standard model higgs boson . the dynamics of the extreme limit of a walking technicolor theory will be responsible for a fermionic self - energy that is given by the limit @xmath18 in eq.([sig ] ) . to compare the trilinear higgs coupling for fundamental and composite scalar bosons we will consider technicolor models with technifermions in the fundamental representation and will choose appropriately the number ( @xmath34 ) of technifermions in order to obtain the desired walking behavior . for example , if the technicolor group is @xmath67 , the walking limit is going to be obtained with @xmath68 . the 8 technifermions can be reconized as a colored weak doublet @xmath69 is a color index . ] , and a color - singlet weak doublet @xmath70 . if the technicolor theory is described by the @xmath71 non - abelian group , the extreme walking behavior is obtained when @xmath72 , which can be built with the addition of two colored weak singlets @xmath73 . of course we are not discussing about phenomenologically viable models , but the cases that we are presenting are plausible examples to make the comparison between the composite " and the elementary coupling . in the fig.(2 ) the continuous curve shows the behavior of the trilinear higgs boson self - coupling given by eq.([e18 ] ) . in the same figure we indicate by @xmath74 the values of the trilinear composite higgs couplings obtained respectively with the help of eqs.([e19 ] ) and ( [ e20 ] ) @xmath75 in the case of the @xmath67 technicolor group . we also indicate by @xmath76 in fig.(2 ) the values of the trilinear coupling obtained for the @xmath71 when the parameter @xmath38 has respectively the following behavior @xmath75 . it is possible to verify in fig.(2 ) that the trilinear higgs coupling generated by the dynamics in the limit @xmath18 , which corresponds to the extreme walking technicolor limit , are quite close to the values obtained in the case of the fundamental standard model higgs boson . however , in the limit @xmath17 the behavior predicted for the trilinear higgs coupling is very different ; it decreases the more the technicolor dynamics approaches the standard result predicted by simple ope analysis . the arrow in fig.(2 ) shows roughly the expected change in the trilinear coupling as we go from @xmath77 to @xmath17 . as will be shown in the next section this behavior will imply in smaller cross section for higgs boson pair production . the large number of gluons in high - energy proton beams imply that the gluon - gluon fusion mechanism is the dominant process for higgs boson pair prodution at the lhc@xcite . the gluon - gluon mechanism involves the triangular and box loops of the heavy quarks , as shown in fig.(3 ) , and is probably the best way to measure the trilinear higgs coupling @xcite . in the standard model , only the top quark , and to a lesser extent the bottom quark , will contribute to the amplitudes of the diagrams in fig.(3 ) . in this section we will compute the twin higgs boson production with the trilinear couplings that we obtained in the previous sections , in order to verify what this process can tell to us about the structure of the technicolor theory . in terms of the trilinear higgs coupling , @xmath78 , the elementary @xmath79 cross section at leading order can be written as@xcite = ^^+_^-d where the couplings @xmath80 and @xmath81 are defined as @xmath82 with @xmath83 , @xmath84 $ ] and _ h = . the form factors @xmath85 , @xmath86 and @xmath87 come from the triangle and box loop evaluation . it is possible to derive simple expressions for the cross section in the limit where the higgs boson mass is much lighter or much heavier than the internal quark ( @xmath88 ) mass that is running in the loops . according to the discussion of the previous section it is possible to assume the limit where @xmath89 , because for @xmath90 we can expect that @xmath91@xcite . in this case the form factors @xmath92 and @xmath93 take a very simple form according to the last article quoted in the ref.@xcite f__t ~-,f__b ~g__b 0 , what allow us to write the partonic cross section as = ^2__hhh\|f__t\|^2 . the total cross section for higgs boson pairs production through gluon - gluon fusion in pp collisions can be determined by integrating over the @xmath94 luminosity@xcite _ t(pp hh ) = ^1_4m^2__h / sd(=s ) where = ^1_g(x,^2)g(/x,^2 ) and we use the gluons distribution function ( @xmath95 ) obtained in ref.@xcite , taken at a typical scale @xmath96 . the cross section contains terms proportional to @xmath97 ( where @xmath98 is the mass of the quark running in the loop ) which were expanded and only the leading term was considered . this is reasonable because the heaviest quark mass is the one of the top quark and we are considering heavier scalar masses . these equations have been numerically integrated and in the sequence we plot the cross - section as a function of the higgs boson mass . in fig.(4 ) the curve shows the standard model prediction for higgs boson pair production . we are also pointing out four values of the cross - section for the cases of composite higgs boson that were discussed in the previous section . the points indicated by ( @xmath99 ) are the predictions for the @xmath67 technicolor group respectively in the limits @xmath64 and @xmath65 . the composite higgs boson mass is estimated as being of order @xmath100 as we vary the value of @xmath38 in the range @xmath101 . in the same figure we are also considering the case where @xmath71 is the tc group , again its predictions are indicated by ( @xmath102 ) . the higgs mass obtained in this case is @xmath103 . to compute these cross sections we used the top quark mass equal to @xmath104 . the cross section falls with increasing @xmath61 due to both the diminishing luminosity and the decrease of the phase space . according to what we commented in the end of sec.iii in the case of composite higgs boson the cross section decreases when @xmath17 ( see the arrow depicted in the fig.4 ) this fact is a consequence of a combination of two effects in the cross section : decreasing of @xmath105 due to the increase in @xmath38 ( see fig.(2 ) ) , and the increase of @xmath106 . although enhancement of the signal for higgs boson pair production is possible for light standard model scalar bosons@xcite , and even in some extensions of the standard model@xcite , we do not see any possible enhancement of the trilinear self - coupling , and , consequently , of the higgs boson pair production in the case of heavy dynamically generated scalar bosons . note that the cross section will be basically impossible to be measured at the lhc , even if the self - couplings , determined from eqs.([e19 ] ) and ( [ e20 ] ) , are of @xmath107 for the scalar masses that we are considering . there are possible refinements in these calculations , like the introduction of nlo corrections@xcite , which are not considered because we just want an order of magnitude estimate , therefore any refinement that may be introduced will not modify the fact that the production of heavy composite higgs bosons will have smaller rates than in the case of fundamental bosons . in this work we have presented a discussion about the general properties of the trilinear self - coupling of a composite higgs boson based on a general ansatz for the technifermion self - energy . if the higgs boson is composite we can expect it to be , at least in the most usual models , a very massive particle , @xmath108 , as in the examples of technicolor gauge groups discussed above ( @xmath71 or @xmath67 ) . we verified in the section iv , assuming in our ansatz the extreme walking technicolor limit , @xmath18 , that the cross section for higgs boson pair production depicted in the fig.(4 ) practically does not differ from the one of the fundamental standard model higgs boson . the cross section in the limit @xmath17 is very different from the standard result . according to what is known for a long time , this dynamics is exactly the one that leads to the many phenomenological problems in technicolor models . in conclusion , if the origin of the gauge symmetry breaking of the standard model is a dynamical one , _ i.e. _ the scalar higgs boson is a composite one , it will be much harder than it is in the case of the standard model ( with fundamental scalar bosons ) to obtain information about the mechanism with the study of the trilinear self - couplings at the lhc . the best possibility to study this coupling , at least for the models usually discussed in the literature , will happens in the case of an extreme walking technicolor dynamics . this research was supported by the conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) ( aan ) and by fundaco de amparo pesquisa do estado de so paulo ( fapesp ) ( ad ) . 99 j. f. gunion et al . , the higgs hunter s guide " , ( frontiers in physics ) , addison - wesley , reading , ma , 1990 . u. baur , t. plehn and d. l. rainwater , _ phys . lett . * 89 * _ , 151801 ( 2002 ) ; u. baur , t. plehn and d. l. rainwater , _ phys . rev . * d67 * _ , 033003(2003 ) ; a. djouadi , hep - ph/0503172 . k. lane , _ technicolor 2000 _ , lectures at the lnf spring school in nuclear , subnuclear and astroparticle physics , frascati ( rome ) , italy , may 15 - 20 , 2000 ; hep - ph/0007304 ; see also hep - ph/0202255 ; r. s. chivukula , _ models of electroweak symmetry breaking _ , nato advanced study institute on quantum field theory perspective and prospective , les houches , france , 16 - 26 june 1998 , hep - ph/9803219 . a. djouadi , w. kilian , m. muhlleitner and p. m. zerwas , hep - ph/0001169 ; a. djouadi , w. kilian , m. muhlleitner and p. m. zerwas , _ eur . phys . j. _ * c10 * , 45 ( 1999 ) ; m. spira , a. djouadi , d. graudenz and p. m. zerwas , _ nucl . phys . * b453 * _ , 17 - 82(1995 ) ; t. plehn , m. spira and p.m. zerwas , _ nucl.phys . _ * b479 * , 46 ( 1996 ) ; _ erratum - ibid _ , * b531 * , 655 ( 1998 ) . c. t. hill and e. h. simmons , _ phys . rept . _ * 381 * , 235 ( 2003 ) [ erratum - ibid . * 390 * , 553 ( 2004 ) ] . b. holdom , _ phys . rev . _ * d24 * , r1441 ( 1981);_phys . lett . _ * b150 * , 301 ( 1985 ) ; t. appelquist , d. karabali and l. c. r. wijewardhana , _ phys . rev . lett . _ * 57 * , 957 ( 1986 ) ; t. appelquist and l. c. r. wijewardhana , _ phys . rev . _ * d36 * , 568 ( 1987 ) ; k. yamawaki , m. bando and k.i . matumoto , _ phys . lett . _ * 56 * , 1335 ( 1986 ) ; t. akiba and t. yanagida , _ phys . * b169 * , 432 ( 1986 ) . d. d. dietrich , f. sannino and k. tuominen , _ phys . _ * d72 * , 055001 ( 2005 ) . a. doff and a. a. natale , _ phys . _ * b537 * , 275 ( 2002 ) . j. carpenter , r. norton , s. siegemund - broka and a. soni , _ phys . _ * 65 * , 153 ( 1990 ) . j. d. carpenter , r. e. norton and a. soni , _ phys . * b 212 * , 63 ( 1988 ) . r. delbourgo and m. d. scadron , _ phys . lett . _ * 48 * , 379 ( 1982 ) . a. doff and a. a. natale , _ phys . rev . _ * d68 * , 077702 ( 2003 ) ; _ eur . phys . j. _ * c32 * , 417 ( 2004 ) . r. alkofer and l. von smekal , _ phys . rep . _ * 353 * , 281 ( 2001 ) ; a. c. aguilar , a. mihara and a. a. natale , _ phys . rev . _ * d65 * , 054011 ( 2002 ) ; _ int . phys . _ * a19 * , 249 ( 2004 ) , a. c. aguilar and a. a. natale , jhep 0408 , 057 ( 2004 ) . a. c. aguilar , a. a. natale and p. s. rodrigues da silva , _ phys . lett . _ * 90 * , 152001 ( 2003 ) . s. j. brodsky , hep - ph/0111127 ; acta phys . * b32 * , 4013 ( 2001 ) ; fortsch . phys . * 50 * , 503 ( 2002 ) ; hep - ph/0310289 . h. pagels and s. stokar , _ phys . _ * d20 * , 2947 ( 1979 ) . s. raby , s. dimopoulos and l. susskind , _ nucl . phys . _ * b169 * , 373 ( 1980 ) . o. j. p. eboli , g. c. marques , s. f. novaes and a. a. natale , _ phys . lett . _ * b197 * , 269 ( 1987 ) . we are considering the most usual models where the higgs boson turns out to be quite heavy . in ref.@xcite it was shown that the walking behavior can be obtained by introducing technifermions in higher dimensional representations of the technicolor gauge group with a very small number of fields , and also obtaining a very light mass for the higgs boson , i. e. @xmath109 . a. d. martin , w. j. stirling and r.g . roberts , _ phys * b354 * , 155 ( 1995 ) . s. dawson , s. dittmaier and m. spira , _ phys . * d 58 * , 115012(1998 ) .	we discuss the strength of the trilinear higgs boson coupling in technicolor ( or composite ) models in a model independent way . the coupling is determined as a function of a very general ansatz for the technicolor self - energy , and turns out to be equal or smaller than the one of the standard model higgs boson depending on the dynamics of the theory . with this trilinear coupling we estimate the cross section for higgs boson pair production at the lhc . this measurement is quite improbable in the case of a heavy standard model higgs boson , but it will be even worse when this boson is dynamically generated . @xmath0 )
+ + chiral de rham complex of beilinson and drinfeld @xcite is an attempt to extend conformal field theory ( cft ) to the case of target manifolds which do not necessarily obey the condition of the vanishing first chern class of tangent bundle @xmath1 . we usually exclude such a situation from our consideration since non - linear sigma model on such a target manifold contains a logarithmic divergence and hence a non - vanishing beta function . scale invariance will be lost in such a situation and one can not apply the method of cft . one way of realizing the idea of beilinson and drinfeld is to use the patchwise construction by malikov - schectman - vaintrob where one considers cft defined on each coordinate patch and then considers the consistency of the theory under coordinate transformation among different patches . in the case when the manifold is covered by , say , 4 patches @xmath2 , one first considers successive transformations @xmath3 . in the end one finds that under the total coordinate change @xmath4 the fields do not quite come back to their original values but there appears an additional term . namely , there exists an obstruction or anomaly for a consistent cft in such a system . it has been suggested that the obstruction is related to the first potryagin class of the manifold [ gorbounov - malikov - schechtman , witten , nekrasov , et.al . ] . in this paper we follow the approach of malikov - schechtman - vaintrob@xcite and study the case of del pezzo surfaces which are rational surfaces obtained from @xmath0 by blowing up a certain number of points . we study the cases of one - point , two - point , and three - point blowups of @xmath0 and show that the anomaly on these surfaces is in fact proportional to the first pontrjagin class . we consider this is a substantial evidence in favor of the suggestion by witten and nekrasov . in chapter 2 of this these we start with the general theory of @xmath5 system ( conformal dimensions of @xmath6 are 0 and 1 , respectively ) where @xmath7 field is identified as the local coordinate of the manifold and @xmath8 field is identified an 1-form . following [ malikov - schechtman - vaintrob ] and nekrasov we discuss the transformation laws of @xmath6 system under coordinate change so that their ope is preserved . then in chapter 3 we discuss the case of @xmath0 as the target manifold and reproduce the result of witten . in chapter 4 we consider the case of del pezzo surface which is an 1,2,3-point blow up of @xmath0 , which has the vanishing pontrjagin anomaly in the 1 point case . we compute the obstruction for these cases and find that they are proportional to the first pontrjagin class of the manifolds . computations are somewhat lengthy but straightforward . non - trivial aspect of the computation is the change in the convention of the normal ordering when one goes to a different coordinate patch and we have to make a careful analysis . in chapter 5 we present some discussions and conclusions . in appendix a we explain the historical background of wess - zumino - witten theory . this theory is used in the ansatz of nekrasov([nekrasovansatz ] ) as the 1-form @xmath9 . in appendix b we include some brief illustrations of the toric diagrams and blowups . del pezzo surfaces are defined as the complex algebraic surfaces x such that the anticanonical divisor is ample . the anticanonical divisor of x is @xmath10 where @xmath11 is the hyperplane class and @xmath12 is the exceptional divisor of @xmath13 generic point(s ) blowups of @xmath0 . therefore we get @xmath14 . we will treat the generic @xmath15 @xmath16 points blowups of @xmath0 by attaching @xmath17 instead of points . it is differential geometrically , complex algebraic surfaces with positive curvatures . we will check the curved target space of @xmath18 cft , which was examined by witten@xcite in the case of @xmath0 . with a minus factor to the original paper]if we define the commutation relation , @xmath19_- & = & \delta_i^j \delta_{m , -n } c \\ \gamma^i ( z ) & = & \sum_{n \in \mathbb{z } } \gamma^i_n z^{-n } \\ \beta_i ( z ) & = & \sum_{n \in \mathbb{z } } \beta_{i \ n } z^{-n-1},\end{aligned}\ ] ] where @xmath20 is a constant that will be mapped to 1 later in the polynomial ring . @xmath7 has conformal dimension 0 , and @xmath8 has 1 . namely , @xmath21 are the annihilation operators , and @xmath22 are the creation operators . and @xmath23 are the annihilation operators , and @xmath24 are the creation operators . we will regard @xmath25 as the coordinate , and @xmath26 are vector fields . for @xmath27 , or @xmath28 and @xmath29 ( @xmath30 is the state space that the heisenberg algebras act on ) , the normal ordered product @xmath31 is giben by @xmath32 the ope of @xmath8 and @xmath7 is computed as follows @xmath33_- z^{-n-1 } w^{-m } + \gamma^j_m \beta_{i \ n } z^{-n-1 } w^{-m } ) \nonumber\\ & = & \sum_{n < 0 , m } \beta_{i \ n } \gamma^j_m z^{-n-1 } w^{-m } + \sum_{n \ge 0 , m } ( \delta_i^j \delta_{n ,- m } z^{-n-1 } w^{-m } + \gamma^j_m \beta_{i \ n } z^{-n-1 } w^{-m } ) \nonumber\\ & = & \sum_{n < 0 , m } \beta_{i \ n } \gamma^j_m z^{-n-1 } w^{-m } + \sum_{m \ge 0 } \delta_i^j \frac{1}{z - w } + \sum_{n \ge 0 , m } \gamma^j_m \beta_{i \ n } z^{-n-1 } w^{-m } \nonumber\\ & = & \frac{\delta_i^j}{z - w } + : \beta_i ( z ) \gamma^j ( w):.\end{aligned}\ ] ] here it is assumed @xmath34 where the summation of the second term is absolutely convergent , and the first and the third term can be summarized as the normal ordering . we will call this the radial ordering in the following . @xmath35 similarly , @xmath36 the stress energy tensor is given by @xmath37 ope of the stress - energy tensor is given by @xmath38 this can be verified by using the wick theorem . .]now we will consider the fermion fields @xmath39 . the conformal dimension of @xmath40 and @xmath41 are 0 and 1 , respectively . opes of @xmath42-@xmath43 system are given by @xmath44 in order to describe the topological vertex algebra , we introduce the following fields @xmath45 : stress - energy tensor of conformal dimension 2 , @xmath46 : u(1 ) current of conformal dimension 1 , and @xmath47 are ( twisted version of ) two generators of @xmath48 supersymmetry of conformal dimension 1 and 2 , respectively . they are written as follows : @xmath49 \\ j & = & \sum_i : b^i ( z ) c_i ( z ) : \\ q & = & \sum_i : \beta^i ( z ) b^i ( z ) : \\ g & = & \sum_i : c_i(z ) \partial \gamma^i ( z):.\end{aligned}\ ] ] the opes are given by @xmath50 here @xmath51 denotes the dimension of the target manifolds . note that the operator @xmath52 can be expressed as the commutator . this is derived by integrating the equation ( [ qg_ope ] ) around w and we obtain @xmath53_+ & = & l(w)\end{aligned}\ ] ] where @xmath54 is the contour integration around the origin of @xmath55 . since the stress tensor @xmath52 becomes the brst exact , we obtain a topological theory . this algebra is determined by the twist of @xmath48 scft , whose stress energy tensor is @xmath56 . . ] the indices are important in this subsection . we will regard @xmath25 as coordinates of the target space , @xmath26 and @xmath57 as 1-form , and @xmath40 as vector field . when we will change the coordinate @xmath25 to @xmath58 , we have to determine the transformation of the other fields in such a way that they preserve the ope . the result is given by @xmath59 second term of @xmath60 ( [ tilde_beta ] ) is the quantum effect . as we can easily verify it , the ope of @xmath42-@xmath18 system is preserved . the purpose of this section is to derive useful formulae which will be used in the next section . in the next section we want to calculate the ope for @xmath61 where the @xmath62 and @xmath63 are the functions of @xmath7 . we define @xmath64 and @xmath65 for every 1-form @xmath66 , and every vector @xmath67 , where @xmath68 is the coordinate patch which we are working with @xmath69 \\ c_b & = & b_i ( \gamma ( z ) ) \partial \gamma^i.\end{aligned}\ ] ] note that @xmath65 has the conformal weight 1 , whereas @xmath64 has also conformal dimension 1 but has an extra @xmath70 term in the ope with the stress energy tensor @xmath52 . by utilizing the inner product , the lie derivatives , and the commutation relation in the basis of tangent bundle @xmath71 , we obtain @xmath72^j ( z ) & = & \partial_i v^j_b ( z ) v^i_a ( z ) - \partial_i v^j_a ( z ) v^i_b ( z ) \\ \mathcal{l}_v b ( z ) & = & \partial_i b_j ( z ) v^i ( z ) + \partial_j v^i ( z ) b_i ( z).\end{aligned}\ ] ] we can compute the following opes @xmath73 ( z)}}{\epsilon } - \frac{c_{\omega_{ab}}(z)}{\epsilon } \label{jj_ope } \\ j_v ( z + \epsilon ) c_b ( z ) & \sim & \frac{\iota_v b(z)}{\epsilon^2 } + \frac{c_{\mathcal{l}_v b ( z)}}{\epsilon } \label{jc_ope } \\ c_b ( z + \epsilon ) c_{b ' } ( z ) & \sim & 0.\end{aligned}\ ] ] here @xmath74 where the matrix is defined by @xmath75 we will use these equations in the following sections . let us introduce an ansatz of nekrasov@xcite for a coordinate change @xmath25 to @xmath86 such that the ope of the @xmath18-system is unchanged after the coordinate transformation . we write the ansatz like as follows @xmath87 where @xmath88 . then by requiring @xmath89 we obtain @xmath90 now we examine the ope relation @xmath91 by using the ope of ( [ lhs2 ] ) of generalized complex manifolds of @xmath92 we obtain @xmath93 } - c_{\omega_{ab } } ) . \label{beta_beta}\end{aligned}\ ] ] by requiring ( [ betabetaope ] ) , the @xmath94 and @xmath82 terms have to vanish . as for @xmath82 part , we have @xmath95 we now want to determine @xmath96 . let us define the symmetric @xmath97 and antisymmetric @xmath98 part of @xmath99 as follows @xmath100 then , from the definition of symmetric part and ( [ sigma_b ] ) , we can first conclude that @xmath101 we still have to determine @xmath98 . from the @xmath94 part of @xmath102 ope ( [ beta_beta ] ) , we obtain the necessary equation @xmath103 } - c_{\omega_{ab } } & = & j_{[g_a , g_b ] } + c_{\mathcal{l}_{g_a } b_b - \mathcal{l}_{g_b } b_a - \frac12 d \mu_{ab } } - c_{\omega_{ab } } \nonumber\\ & = & c_{\mathcal{l}_{g_a } b_b - \mathcal{l}_{g_b } b_a - \frac12 d \mu_{ab } } - c_{\omega_{ab } } \\ & = & 0.\end{aligned}\ ] ] we contract this equation with any vector @xmath104 to obtain the equation @xmath105 by using the maurer - cartan equation for @xmath106 , this equation can be rewritten as @xmath107 where @xmath108 and there are no further conditions for @xmath109 besides this equation . , width=188 ] let s consider @xmath0 as the target space . we will review the computation of the @xmath110 anomaly of @xcite . let @xmath111 be the affine coordinate patch defined by @xmath112 ( @xmath113 ) for projective coordinate @xmath114 , then , for the coordinate @xmath25 ( @xmath115 ) of patch @xmath116 , we have the ope of ( [ betagamma_ope ] ) @xmath117 under the combined coordinate transformations from @xmath118 to @xmath118 : @xmath119 @xmath120 where @xmath121 is an antisymmetric tensor . then , @xmath122 we will denote @xmath123 as @xmath124 , @xmath125 and @xmath126 as @xmath127 , @xmath128 . we assume that the generator of @xmath129 is given by @xmath130 where @xmath124 , @xmath125 are the inhomogeneous standard coordinate of @xmath131 . @xmath132 then we perform successive coordinate transformations @xmath133 and back to @xmath118 . we first check the coordinate transformation of @xmath134 , which can be read from the toric diagram ( figure [ cp2 ] ) . [ [ section ] ] in the following subsections , as @xcite did , we will use the notation @xmath135 } , w^{[i ] } , v^{[i ] } , w^{[i]}$ ] , instead of @xmath136 . to be more precise , let @xmath137 } & = & \frac{\lambda_{i+1}}{\lambda_i } \\ w^{[i ] } & = & \frac{\lambda_{i+2}}{\lambda_i}\end{aligned}\ ] ] @xmath138}$ ] are the corresponding 1-form . @xmath139 } & = & \frac{w}{v } \\ w^{[1 ] } & = & \frac{1}{v } \\ v^{[1 ] } & = & v w \label{v1}\\ w^{[1 ] } & = & -v^2 v - vw w - \frac{5}{2 } \label{w1 } \partial v.\end{aligned}\ ] ] the jacobian matrix @xmath140 is given by @xmath141 then the inverse matrix is < @xmath142 this gives the jacobian part @xmath143 of nekrasov s formula ( [ nekrasovansatz ] ) that assumes the generalized complex structure . now that we have to determine the @xmath63 of @xmath144 we decompose @xmath9 as @xmath145 , where @xmath146 since @xmath0 is 2 dimensional . on the other hand , the symmetric part is given by @xmath147 for @xmath148}$ ] , @xmath149 for @xmath150}$ ] and @xmath151}$ ] , @xmath152 finally , @xmath153}$ ] , @xmath154 then , @xmath155 [ [ section-1 ] ] similar discussions yield that the coordinate transformations for @xmath156 @xmath157 } & = & \frac{w^{[1]}}{v^{[1 ] } } = \frac{1}{w } \\ w^{[2 ] } & = & \frac{1}{v^{[1 ] } } = \frac{v}{w } \\ v^{[2 ] } & = & v^{[1 ] } w^{[1 ] } \label{v2 } \\ w^{[2 ] } & = & -(v^{[1]})^2 v^{[1 ] } - v^{[1 ] } w^{[1 ] } w^{[1 ] } - \frac{5}{2 } \partial v^{[1 ] } , \label{w2}\end{aligned}\ ] ] and @xmath158 @xmath159 } & = & \frac{w^{[2]}}{v^{[2 ] } } = v\\ w^{[3 ] } & = & \frac{1}{v^{[2 ] } } = w \\ v^{[3 ] } & = & v^{[2 ] } w^{[2 ] } \\ w^{[3 ] } & = & -(v^{[2]})^2 v^{[2 ] } - v^{[2 ] } w^{[2 ] } w^{[2 ] } - \frac{5}{2 } \partial v^{[2]}\end{aligned}\ ] ] to be more careful , when we substitute and combine the equations ( [ v1])([w1 ] ) ( @xmath134)in the equations of @xmath160}$ ] ( [ v2 ] ) and @xmath161}$ ] ( [ w2 ] ) for @xmath156 , which have the cross terms with @xmath162 or @xmath163 . since the definition of normal product depends on the patches and each term of ( [ w2 ] ) is defined by normal product , the first term of @xmath161}$ ] , for example , is defined as @xmath164})^2 v^{[1 ] } & = & \lim_{z ' \to z } ( -v^{[1 ] } ( z')^2 v^{[1]}(z ) + 2 v^{[1]}(z ' ) \frac{1}{z ' - z } ) \nonumber\\ & = & \lim_{z ' \to z } \left [ - ( \frac{w(z')}{v(z')})^2 v(z ) w ( z ) + 2 ( \frac{w(z')}{v(z ' ) } ) \frac{-1}{z ' - z } ) \right ] \nonumber\\ & = & \lim_{z ' \to z } \left [ - \frac{2 w ( z')}{z - z ' } \frac{v(z)}{v(z')^2 } - ( \frac{w(z')}{v(z')})^2 v(z ) w(z ) + 2 ( \frac{w(z')}{v(z ' ) } ) \frac{-1}{z ' - z } \right ] \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ ( + \frac{2}{\epsilon } \frac{w}{v } - 4 \frac{w \partial v}{v^2 } + \frac{2 \partial w}{v } ) - \frac{w^2}{v}w + ( \frac{-2(\partial w v - \partial v w)}{v^2 } - \frac{2}{\epsilon } \frac{w}{v } ) ) \right ] \nonumber\\ & = & -2 \frac{\partial v w}{v^2 } - \frac{w^2}{v } w\end{aligned}\ ] ] by applying similar treatments for the 2nd term of ( [ w2 ] ) , we obtained the relation for @xmath133 . therefore @xmath165 } & = & -vwv -w^2 w - \frac{3 w \partial v}{2v } - \partial w \\ w^{[2 ] } & = & wv - \frac{3 \partial w}{2v}\end{aligned}\ ] ] [ [ section-2 ] ] again , we will substitute the @xmath166 } , w^{[3]}$ ] by the equations above to get @xmath167 } & = & v^{[2 ] } w^{[2 ] } \nonumber\\ & = & \frac{1}{w } : w v - \frac{3}{2 } \frac{\partial w}{v } : \nonumber\\ & = & v - \frac{3}{2 } \frac{\partial w}{v w},\end{aligned}\ ] ] and for @xmath168 @xmath169 } & = & - : ( v^{[2]})^2 v^{[2 ] } : - : v^{[2 ] } w^{[2 ] } w^{[2 ] } : -\frac{5}{2 } \partial v^{[2]}\nonumber\\ & = & - \lim_{\epsilon \to 0 } \left [ ( v^{[2 ] } ( z'))^2 v^{[2 ] } ( z ) + 2 v^{[2 ] } ( z ) \frac{1}{\epsilon } + 2 \partial v^{[2 ] } \right ] \nonumber\\ & & - \lim_{\epsilon \to 0 } \left [ v^{[2 ] } ( z ' ) w^{[2 ] } ( z ' ) w^{[2 ] } ( z ) + v^{[2 ] } \frac{1}{\epsilon } + \partial v^{[2 ] } \right ] \nonumber\\ & & -\frac{5}{2 } \partial v^{[2 ] } \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ - \frac{3}{\epsilon } v^{[2 ] } ( z ) - \frac{11}{2 } \partial v^{[2 ] } - ( v^{[2 ] } ( z'))^2 v^{[2 ] } ( z ) - v^{[2 ] } ( z ' ) w^{[2 ] } ( z ' ) w^{[2 ] } ( z ) \right ] \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ -\frac{5}{2 } \partial v^{[2 ] } + \frac{1}{w(z')^2 } ( vwv + w^2w + \frac{3 w \partial v}{2v } + \partial w ) ( z ) \right ] \nonumber\\ & & + \lim_{\epsilon \to 0 } \left [ - \frac{1}{w(z')}\frac{v(z')}{w(z ' ) } ( w v - \frac{3}{2 } \frac{\partial w}{v } ) ( z ) \right ] \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ - \frac{3}{\epsilon } v^{[2 ] } ( z ) - \frac{11}{2 } \partial v^{[2 ] } + \frac{v v}{w } + \frac{2 w^{-3 } ( z ' ) w^2(z)}{\epsilon } + \frac{3 \partial v}{2 v w } \right ] \nonumber\\ & & + \lim_{\epsilon \to 0 } \left [ \frac{\partial w}{w^2 } + \frac{w ( z)}{w(z')^2 } \frac{1}{\epsilon } - \frac{v v}{w } + \frac{3 \partial w}{2 w^2 } + w(z ) \right ] \nonumber\\ & = & w(z ) + \frac{3}{2 } \frac{\partial v}{vw}\end{aligned}\ ] ] thereby , in total , @xmath170 this certainly is the ansatz of witten ( [ wittenansatz ] ) . we will check whether this result can be derived from the ansatz of nekrasov ( [ nekrasovansatz ] ) with the @xmath171 whereas the total gerbe term @xmath172 is antisymmetric . this is due to the anomaly 2-form of nekrasov . for @xmath174 : affine patches , we perform successive coordinate transformations @xmath175 . the coordinate transformations in each step are given by @xmath176 we will use the indices @xmath177 as @xmath178 by combining these three coordinate transformations , we obtain @xmath179 @xmath173 is defined by @xmath180 where @xmath181 , @xmath182 , and @xmath183 are the antisymmetric parts of @xmath9 , @xmath184 , and @xmath185 . this is called the anomaly 2-form . ( [ btot ] ) can be proved as follows ; for 2 step coordinate changes @xmath186 , @xmath187 \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ g^a_{\alpha } ( z + \epsilon ) ( g^i_a ( z ) \beta_i ( z ) + b_{ai } ( z ) \partial \gamma^i ( z ) ) + \frac{1}{\epsilon } \partial_a g^a_{\alpha } ( z + \epsilon ) \right ] \nonumber\\ & = & \lim_{\epsilon \to 0 } \left [ g^a_{\alpha } g^i_a \beta_i + g^a_{\alpha } b_{ai } ( z ) \partial \gamma^i ( z ) - \frac{1}{\epsilon } \partial_i g^a_{\alpha } ( z + \epsilon ) g^i_a ( z ) + \frac{1}{\epsilon } \partial_a g^a_{\alpha } ( z + \epsilon ) \right ] . \nonumber\\\end{aligned}\ ] ] this can be rewritten as follows by using @xmath188 and @xmath189 , @xmath190 where @xmath191 is given by @xmath192 the 3 step coordinate change @xmath175 ( [ btot ] ) is obtained by combining ( [ 2step ] ) and the ( [ 3rdstep ] ) . let us confirm this reproduce the previous result for the case of @xmath0 . @xmath171 and we can set @xmath109 as arbitrary value . we have to check whether @xmath193 for some @xmath109 after computing @xmath194 @xmath195 } \\ - ( v^{[1]})^2 & -v^{[1 ] } w^{[1 ] } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } 0 & \frac{w}{v } \\ - \frac{w^2}{v^2 } & - \frac{w}{v^2 } \end{array } \right ) \nonumber\\ ( g''^{\alpha}_i)_{i \alpha } & = & \left ( \begin{array}{cc } 0 & v^{[2 ] } \\ - ( v^{[2]})^2 & -v^{[2 ] } w^{[2 ] } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } 0 & \frac{1}{w } \\ - \frac{1}{w^2 } & - \frac{v}{w^2 } \end{array } \right)\end{aligned}\ ] ] therefore @xmath196 and @xmath197 this agrees with the result of direct ope computation in ( [ vwcyclic ] ) . and the anomaly exists because @xmath109 can not eliminate this term . the coordinate change can be read from the toric diagram ( figure [ toric1 ] ) and ( figure [ del1 ] ) , width=188 ] , width=377 ] @xmath198 } & = & x = ( \gamma_1^{[3]})^{-1 } \nonumber\\ \gamma_2^{[1 ] } & = & y = ( \gamma_1^{[3]})^{-1 } ( \gamma_2^{[3]})^{-1 } \nonumber\\ \gamma_1^{[2 ] } & = & x^{-1 } \nonumber\\ \gamma_2^{[2 ] } & = & x^{-1 } y \nonumber\\ \gamma_1^{[3 ] } & = & x^{-1 } = \gamma_1^{[2 ] } \nonumber\\ \gamma_2^{[3 ] } & = & x y^{-1 } = ( \gamma_2^{[2]})^{-1}\end{aligned}\ ] ] the jacobian matrices ( note that we are not dealing with inverse jacobians ) for @xmath199 are given by @xmath200}}{\partial \gamma_i^{[1 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } -x^{-2 } & - x^{-2 } y \\ 0 & x^{-1 } \end{array } \right ) \nonumber\\ ( g'^{\alpha}_a)_{a \alpha } & = & \frac{\partial \gamma_{\alpha}^{[3]}}{\partial \gamma_a^{[2 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - x^2 y^{-2 } \end{array } \right ) \nonumber\\ ( g''^i_{\alpha})_{\alpha i } & = & \frac{\partial \gamma_i^{[1]}}{\partial \gamma_{\alpha}^{[3 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } -x^2 & - x y \\ 0 & - x^{-1 } y^2 \end{array } \right)\end{aligned}\ ] ] then @xmath201 therefore @xmath202 \nonumber\\ & = & dx \wedge dy \left ( \begin{array}{cc } 0 & 2 x^{-1 } y^{-2 } \\ 0 & -2 y^{-3 } \end{array } \right ) \nonumber\\ \psi^0_{123 } = tr g '' dg \wedge dg ' & = & 2 \frac{dx \wedge dy}{xy}\end{aligned}\ ] ] similarly , for @xmath203 @xmath204}_1 & = & x = \gamma^{[4]}_1 \\ \gamma^{[1]}_2 & = & y = ( \gamma^{[4]}_2)^{-1 } \\ \gamma^{[2]}_1 & = & x^{-1 } \\ \gamma^{[2]}_2 & = & x^{-1 } y \\ \gamma^{[4]}_1 & = & x = ( \gamma^{[2]}_1)^{-1 } \\ \gamma^{[4]}_2 & = & y^{-1 } = \gamma^{[2]}_1 ( \gamma^{[2]}_2)^{-1}\end{aligned}\ ] ] and the jacobians are @xmath205}}{\partial \gamma_i^{[1 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } -x^{-2 } & - x^{-2 } y \\ 0 & x^{-1 } \end{array } \right ) \nonumber\\ ( h'^{\alpha}_a ) & = & \frac{\partial \gamma_{\alpha}^{[4]}}{\partial \gamma_a^{[2 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } - ( \gamma^{[2]}_1)^{-2 } & ( \gamma^{[2]}_2)^{-1 } \\ 0 & - \gamma^{[2]}_1 ( \gamma^{[2]}_2)^{-1 } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } -x^2 & x y^{-1 } \\ 0 & -x y^{-2 } \end{array } \right ) \nonumber\\ ( h''^i_{\alpha } ) & = & \frac{\partial \gamma_{i}^{[1]}}{\partial \gamma_{\alpha}^{[4 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - ( \gamma^{[4]})^{-2 } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - y^2 \end{array } \right)\end{aligned}\ ] ] then @xmath206 therefore @xmath207 \nonumber\\ & = & dx \wedge dy \left ( \begin{array}{cc } 0 & x^{-2 } y^{-2 } \\ 0 & 2 x^{-1 } y^{-3 } \end{array } \right ) \nonumber\\ \psi^0_{124 } = tr h '' dh \wedge dh ' & = & 2 \frac{dx \wedge dy}{xy}.\end{aligned}\ ] ] for @xmath208 @xmath204}_1 & = & x = \gamma^{[4]}_1 \\ \gamma^{[1]}_2 & = & y = ( \gamma^{[4]}_2)^{-1 } \\ \gamma_1^{[3 ] } & = & x^{-1 } \\ \gamma_2^{[3 ] } & = & x y^{-1 } \\ \gamma^{[4]}_1 & = & x = ( \gamma^{[3]}_1)^{-1 } \\ \gamma^{[4]}_2 & = & y^{-1 } = \gamma^{[3]}_1 \gamma^{[3]}_2\end{aligned}\ ] ] and the jacobians are @xmath209}}{\partial \gamma_i^{[1 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } -x^{-2 } & - y^{-1 } \\ 0 & x y^{-2 } \end{array } \right ) \nonumber\\ ( k'^{\alpha}_a ) & = & \frac{\partial \gamma_{\alpha}^{[4]}}{\partial \gamma_a^{[3 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } - ( \gamma^{[3]}_1)^{-2 } & \gamma^{[3]}_2 \\ 0 & \gamma^{[3]}_2 \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } -x^2 & x y^{-1 } \\ 0 & -x^{-1 } \end{array } \right ) \nonumber\\ ( k''^i_{\alpha } ) & = & \frac{\partial \gamma_{i}^{[1]}}{\partial \gamma_{\alpha}^{[4 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - ( \gamma^{[4]}_2)^{-2 } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - y^2 \end{array } \right)\end{aligned}\ ] ] then @xmath210 therefore @xmath211 \nonumber\\ & = & dx \wedge dy \left ( \begin{array}{cc } 0 & -3 x^{-2 } y^{-2 } \\ 0 & 2 x^{-1 } y^{-3 } \end{array } \right ) \nonumber\\ \psi^0_{134 } = tr k '' dk \wedge dk ' & = & -2 \frac{dx \wedge dy}{xy}.\end{aligned}\ ] ] for @xmath212 @xmath213}_1 & = & x^{-1 } = ( \gamma^{[4]}_1)^{-1 } \\ \gamma^{[2]}_2 & = & x^{-1 } y = ( \gamma^{[4]}_1)^{-1 } ( \gamma^{[4]}_2)^{-1}\\ \gamma_1^{[3 ] } & = & x^{-1 } = \gamma^{[2]}_1 \\ \gamma_2^{[3 ] } & = & x y^{-1 } = ( \gamma^{[2]}_2)^{-1 } \\ \gamma^{[4]}_1 & = & x = ( \gamma^{[3]}_1)^{-1 } \\ \gamma^{[4]}_2 & = & y^{-1 } = \gamma^{[3]}_1 \gamma^{[3]}_2\end{aligned}\ ] ] and the jacobians are @xmath214}}{\partial \gamma_i^{[2 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & - ( \gamma^{[2]}_2)^{-2 } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } 1 & 0 \\ 0 & -x^2 y^{-2 } \end{array } \right ) \nonumber\\ ( l'^{\alpha}_a ) & = & \frac{\partial \gamma_{\alpha}^{[4]}}{\partial \gamma_a^{[3 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } - ( \gamma^{[3]}_1)^{-2 } & \gamma^{[3]}_2 \\ 0 & \gamma^{[3]}_2 \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } -x^2 & x y^{-1 } \\ 0 & -x^{-1 } \end{array } \right ) \nonumber\\ ( l''^i_{\alpha } ) & = & \frac{\partial \gamma_{i}^{[2]}}{\partial \gamma_{\alpha}^{[4 ] } } \nonumber\\ & = & \left ( \begin{array}{cc } - ( \gamma^{[4]}_1)^{-2 } & - ( \gamma^{[4]}_1)^{-2 } ( \gamma^{[4]}_2)^{-1 } \\ 0 & - ( \gamma^{[4]}_1)^{-1 } ( \gamma^{[4]}_2)^{-2 } \end{array } \right ) \nonumber\\ & = & \left ( \begin{array}{cc } -x^{-2 } & -x^{-2 } y \\ 0 & - x^{-2 } y^2 \end{array } \right)\end{aligned}\ ] ] then @xmath215 therefore @xmath216 \\ & = & dx \wedge dy \left ( \begin{array}{cc } 0 & 0 \\ 0 & 2 y^{-3 } \end{array } \right ) \nonumber\\ \psi^0_{234 } = tr l '' dl \wedge dl ' & = & -2 \frac{dx \wedge dy}{xy}\end{aligned}\ ] ] in the previous section , we obtained the last term @xmath194 in the anomaly 2-form ( [ psi ] ) as @xmath217 where the coefficients @xmath218 are given by @xmath219 in this section we discuss the other three @xmath109-terms in ( [ psi ] ) . if @xmath194 is cancelled by choosing appropriate @xmath109-terms , the anomaly is absent . this is indeed the case for the 1 point blowup of @xmath0 . this is well described in terms of cech cohomology . @xmath220 where x is the 1 point blowup of @xmath0 and @xmath221 is the chiral de rham complex and @xmath222 is @xmath221 s i - th cech complex and @xmath223 is the corresponding coboundary operator . @xmath194 is described as the 2-cochain @xmath224 . @xmath225 the @xmath109-terms are 1-cochain . here we only consider @xmath181 in the form of @xmath226 @xmath181 must be regular in @xmath227 . by this condition , the variables other than @xmath228 or @xmath229 are zero . this can be shown as follows . we denote the general form of the generator as follows @xmath230 where @xmath231 and @xmath232 are the affine coordinates of @xmath233 . for the 2-form ( [ nm ] ) to be regular in the patch @xmath233 , the integers @xmath15 and @xmath234 must satisfy @xmath235 in order to determing the regularity condition for @xmath236 , we first rewrite the 2-form ( [ nm ] ) with the affine coordinates @xmath237 in the patch @xmath236 . the coordinate transformation between @xmath233 and @xmath236 is given in the figure [ del1 ] . @xmath238 by this coordinate change , the 2-form ( [ nm ] ) is transformed as @xmath239 the regularity in @xmath236 require that the powers of @xmath240 and @xmath241 are non - negative , and we obtain the condition @xmath242 in the same way , the condition for @xmath243 and @xmath244 are obtained as follows @xmath245 these conditions define the allowed region in the lattice ( @xmath15 , @xmath234 ) for each patch . the allowed region for @xmath246 can be obtained by the convex hull of two regions for @xmath247 and @xmath248 . we can easily see that the allowed regions for @xmath249 do not contain @xmath250 , and the corresponding coefficients @xmath251 are zero . the general form of @xmath252 is @xmath253 the condition for anomaly cancellation @xmath254 is rewritten as @xmath255 where @xmath223 is the coboundary operator . this is equivalent to the following relations among coefficients @xmath256 where @xmath257 because these equations have the solution @xmath258 we conclude that @xmath259 therefore the pontjagin anomaly vanishes in this case . from the hirzebruch riemann - roch theorem for surface ( see hartshorne remark 1.6.1 of chapter v ) @xmath260 where @xmath261 is the arithemetic genus of the surface x , which is known to be equal to 0 ( see the theorem 6.2 of hartshorne chapter v by castelnuovo ) for rational surfaces . as the del pezzo surfaces are the special cases of rational surfaces , the lhs = 12 . on the other hand , the anticanonical divisor of x is @xmath10 where h is the hyperplane class and @xmath12 is the exceptional divisor of @xmath13 generic point(s ) blowups of @xmath0 . therefore we get @xmath14 , and the 2nd chern class @xmath262 is @xmath263 now that the 2nd chern character is , keeping in mind that @xmath264 , @xmath265 we conclude that the 2nd chern character vanishes if and only if @xmath266 . we expect that the anomaly 2-form for 1 point blowup should be canceled by the @xmath109-term . , width=188 ] , width=377 ] in general , the total anomaly is given as the sum of @xmath218 for each triangle in a triangulation of the coordinate change diagram ( figure [ del2 ] ) . , width=188 ] , width=377 ] in this section , we discuss the anomaly for the generic 3 points blowup of @xmath0 . in the same way in the previous sections , we obtain the following coefficients @xmath218 of @xmath194 . @xmath280 for the intersections @xmath246 , only the coefficients @xmath281 and @xmath282 are non - vanishing . the total anomaly , the gauge invariant linear combination of @xmath218 , is @xmath283 from the results of section 3.1 , 4.2 , 4.4 , 4.5 , we find that the total anomaly with the proportional constant @xmath284 is @xmath285 for @xmath286 generic points blowup of @xmath0 ( toric rational del pezzo surfaces ) . the 2nd chern character is in agreement with this discussion . in this paper , we explicitly examined by 2 ways that the ansatzes of witten s heterotic ( 0,2 ) model and the nekrasov s generalized complex geometry are consistent . one way is by step by step careful ope calculation and the other is the computation of the anomaly 2-form the 2-cocycle of the chiral de rham complex in terms of coordinate transformation jacobian matrices . we compute the anomaly 2-forms in the case of toric del pezzo surfaces of all degrees and conclude that this coincides ( up to proportional constant 2 ) with the results of hirzebruch riemann - roch theorem . the most notable case is the 1 point blowup , where the total gauge invariant anomaly vanishes . note that beilinson - drinfeld chiral algebra has a background of geometric langlands programs studied on the geometric quantization of vector bundles ( stable higgs bundles ) , which is similar to physicists theory of topological fields or topological strings , and drinfeld s algebro - geometrical construction of wess - zumino - witten model as quantization of 2-dimensional yang - mills theory . the author has also worked in this area for several years and in the near future he would like to summarize this result on the automorphic representation theory with physics interpretation in the spirit of drinfeld , kapranov - vasserot s motivic integration theory . this is a bridge between the twist methods of 4 dimensional super - yang - mills theory of gukov , kapustin - witten et.al . and the local geometric langlands program of edward frenkel , gaitsgory et.al . during the preparation of this paper , kapustin - witten@xcite submitted a paper to the preprint , which is related to this paper . i would like to thank to the support and discussion with my supervisor t.eguchi together with t.terasoma . i cordially thank to y.imamura for his verification of my drafts on the reviews and extensions of malikov - schechtman , nekrasov , and witten . i owe a lot of background mathematic knowledge to d.ben-zvi , h.iritani , and y.shimizu for the semi - infinite homology and the generalized fourier - mukai transformation . i also would like to thank b.dubrovin for the discussion on quantum hitchin systems . i owe the idea of genus 0 path - integral over virtual fundamental class in the case of del pezzo 9 to j.bryan . i thank a.braverman for the instanton counting of semisimple gauge group . i also thank e.frenkel for his comment on frenkel - losev and malikov - schechtman . i acknowledge the discussions with l.hollands , y.konishi , and s.okada for the topological string theory . my work was in part done in the seattle05 summer institute held by ams , the algebraic geometry and beyond held by rims , and the affine hecke algebras , the langlands program , and conformal field theory and matrix models held by cirm . the wess - zumino - witten term is used as a term corresponding to the 1-form @xmath9 , which is used in the nekrasov s ansatz ( [ nekrasovansatz ] ) . we will make a historical note on this term . first , we will think of the @xmath287 spontaneously broken to the diagonal @xmath288 , which involves the goldstone boson @xmath289 . @xmath290 the euler - lagrange equation is @xmath291 we will add a term , which violate @xmath292 that changes @xmath293 , @xmath294 , @xmath295 . @xmath296 we would like to derive this equation from a lagrangian , which is difficult in the first sight . the equation of motion in the constrained system @xmath297 can have both @xmath293 and @xmath298 symmetry , if we write it as follows @xmath299 as the right hand side can be seen as the lorenz force for an electric charge interacting with a magnetic monopole located at the center of the sphere . in this case the vector potential is , by definition , @xmath300 then by the stokes theorem , we can rewrite the contour integral of vector potential as the integration of field strength ( flux ) trough a topological disk @xmath301 , whose choise can be arbitrary and we can especially take as @xmath302 ( orientation reversed ) . @xmath303 therefore @xmath304 is integer or half - integer , which is the dirac quantization . as is noted above , we need some term proportional to the field strength , so we let @xmath305 be five - dimensional disc which has the @xmath288 as the boundary because of @xmath306 @xmath307 @xmath305 can be deformed to orientation revered @xmath308 , then @xmath309 is a five - dimensional sphere . @xmath310 then @xmath311 is in @xmath288 and @xmath312 therefore , we can take a unit sphere @xmath313 and write the action @xmath314 as @xmath315 in the ansatz of nekrasov([nekrasovansatz ] ) in section 2.7 , we derived the symmetric part @xmath316 and antisymmetirc part @xmath109 of the 1-form @xmath9 . if we combine them , we obtain @xmath317 where @xmath318 is the inner derivative . if we rewrite ( [ combinednekrasov ] ) by multiplying @xmath319 from the right and use the cyclicity of trace , we obtain @xmath320 where @xmath321 is the level one wess - zumino - witten lagrangian @xmath322.\end{aligned}\ ] ] toric varieties are algebraic manifolds which have the action of algebraic torus @xmath323 , where n is the complex dimension of the varieties . for each toric diagram ( fan ) , we can define the following data . vertices @xmath324 , and vectors @xmath325 from the origin to the vertices . the dual basis @xmath326 for affine patch @xmath247 spaned by the 2 vectors @xmath327 is such that @xmath328 the affine patch @xmath247 is described by @xmath329,\end{aligned}\ ] ] where @xmath330 , and @xmath331 are the standard bases of @xmath332 . namely the canonical coordinates @xmath135 } , w^{[i]}$ ] in the affine patch @xmath247 are given by @xmath137 } & = & x^{(w_i^1 , e_1)}y^{(w_i^1 , e_2 ) } \\ w^{[i ] } & = & x^{(w_i^2 , e_1 ) } y^{(w_i^2 , e_2)}.\end{aligned}\ ] ] the blowups are operations in the divisor linear system such that we assign a new vertex @xmath333 ( @xmath334 is the number of original vertices ) in the generic point and the corresponding vector from the origin . this operation is birational since we have the toric action on the toric variety , which includes the toric variety @xmath335 as the dense submanifold and therefore the toric diagram in @xmath332 has the @xmath336 action . 100 sergey arkhipov , mikhail kapranov : `` toric arc schemes and quantum cohomology of toric varieties '' , arxiv : math.ag/0410054 dedenis auroux , ludmil katzrkov , and dmitri orlov : `` mirror symmetry for del pezzo surfaces : vanishing cycles and coherent sheaves '' , arxiv : math.ag/0506166 van den bergh , michel : `` blowing up of non - commutative smooth surfaces '' , mem . soc . 154 ( 2001 ) lev a. borisov : `` chiral rings of vertex algebras of mirror symmetry '' , math . z. 248 ( 2004 ) , no . 3 , 567 - 591 , lev a. borisov , anatoly libgober : `` elliptic genera of singular varieties , orbifold elliptic genus and chiral derham complex '' , duke math . j. 116 ( 2003 ) , no . 2 , 319 - 351 d. bump and j.bernstein et.al . : `` an introduction to the langlands program '' , birkhauser ( 2003 ) beilinson , a. ; j. bernstein : `` a proof of jantzen conjectures '' , advances in soviet mathematics vol 16 , part 1 , 1993 , 1 - 50 beilinson , a. a. ; bernstein , j. ; deligne , p. : `` faisceaux pervers '' ( french ) , analysis and topology on singular spaces , i ( luminy , 1981 ) , 5171 , ast@xmath337risque , 100 alexander beilinson , vladimir drinfeld : `` chiral algebras '' , american mathematical society ( 2004 ) beilinson , a. a. ; drinfeld , v. g. : `` quantization of hitchin s fibration and langlands program . '' , algebraic and geometric methods in mathematical physics ( kaciveli , 1993 ) , 37 , math . , 19 , kluwer acad . publ . , dordrecht , 1996 beilinson , a. a. ; drinfeld , v. g. : `` quantization of hitchin s integrable system and hecke eigensheaves '' , available at http://www.math.uchicago.edu/@xmath338arinkin/langlands/ barth , wolf p. ; hulek , klaus ; peters , chris a. m. ; van de ven , antonius : `` compact complex surfaces '' , second edition . ergebnisse der mathematik und ihrer grenzgebiete . 3 . folge . a series of modern surveys in mathematics [ results in mathematics and related areas . 3rd series . a series of modern surveys in mathematics ] , 4 , springer - verlag , berlin , 2004 jim bryan , naichung conan leung : `` the enumerative geometry of k3 surfaces and modular forms '' , arxiv : alg - geom/9711031 a. beilinson and v. schechtman , `` determinant bundles and virasoro algebras '' , commun . . 118 ( 1988 ) 651 - 701 paul bressler , yan soibelman : `` mirror symmetry and deformation quantization '' , arxiv : hep - th/0202128 daniel bump : `` automorphic forms and representations '' , cambridge studies in advanced mathematics 55 alexander beilinson , vladimir drinfeld : `` chiral algebras '' , american mathematical society , dennis gaitsgory : `` notes on 2d conformal field theory and string theory '' , arxiv : math.ag/9811061 , edward frenkel , david ben - zvi : `` vertex algebras and algebraic curves '' , especially chapter 14 ( w - algebra ) , and 16 - 18 ( moduli theory ) , american mathematical society m.demazure : `` surfaces de del pezzo '' , ii , iii , iv et v , in : s@xmath337minaire sur les singularit@xmath337s des surface ( demazure , pinkham , teissier eds . ) , lecture notes in math . 777 , springer - verlag , 23 - 69 ( 1980 ) boris dubrovin , `` geometry of 2d topological field theories '' , integrable systems and quantum groups ( montecatini terme , 1993 ) , 120 - 348 , lecture notes in math . , 1620 , springer , berlin , 1996 boris dubrovin , `` geometry and analytic theory of frobenius manifolds '' , talk at icm98 vladimir drinfeld : `` infinite - dimensional vector bundles in algebraic geometry ( an introduction ) '' , arxiv : math.ag/0309155 , v.a.voevodsky : `` cycles , transfers , and motivic homology theories '' , ann . of math . , 143 , princeton univ . press , princeton , nj , 2000 , ye . a. nisnevich : `` the completely decomposed topology on schemes and associated descent spectral sequences in algebraic @xmath339-theory '' , algebraic @xmath339-theory : connections with geometry and topology ( lake louise , ab , 1987 ) , 241 - 342 j.denef and f.loeser : `` germs of arcs on singular algebraic varieties and motivic integration '' , invent . ( 1999 ) , no . 1 , 201 - 232 edward frenkel , dennis gaitsgory : `` local geometric langlands correspondence and affine kac - moody algebras '' , arxiv : math.rt/0508382 william fulton : `` introduction to toric varieties '' , princeton university press e. frenkel : `` mirror symmetry in two steps : a i b '' , arxiv : hep - th/0505131 e. frenkel : `` lectures on the langlands program and conformal field theory '' , arxiv : hep - th/0512172 e. frenkel : `` ramifications of the geometric langlands program '' , arxiv : math.qa/0611294 e. frenkel : `` langlands correspondence for loop groups '' , cambridge studies in advanced mathematics ( 2007 ) e.frenkel and e.witten : `` geometric endoscopy and mirror symmetry '' , arxiv:0710.5939 v.gorbounov , f.malikov : `` vertex algebras and the landau - ginzburg / calabi - yau correspondence '' , arxiv : math.ag/0308114 , f.malikov , v.schechtman : `` deformations of chiral algebras and quantum cohomology of toric varieties '' , arxiv : math.ag/0001170 l. goettsche , pandharipande , p. : `` the quantum cohomology of blow - ups of @xmath340 and enumerative geometry '' , arxiv : alg - geom/9611012 pietro antonio grassi , giuseppe policastro , and emanuuel scheidegger : `` partition functions , localization , and the chiral de rham complex '' , arxiv : hep - th/0702044 sergei gukov , edward witten : `` gauge theory , ramification , and the geometric langlands program '' , arxiv : hep - th/0612073 nigel hitchin : `` stable bundles and integrable systems '' , duke math . j. 54 ( 1987 ) , 91 - 114 , `` flat connections and geometric quantization '' , comm . math . phys . 131 ( 1990 ) , 347 - 380 , alexander beilinson , vladimir drinfeld : `` quantization of hitchin s integrable system and hecke eigensheaves '' , available at http://www.math.uchicago.edu/@xmath338arinkin/langlands/ , carlos t. simpson : `` higgs bundles and local systems '' , inst . hautes @xmath341tudes sci . publ . math . no . 75 ( 1992 ) , 5 - 95 . gerald t hooft : `` on the phase transition towards permanent quark confinement '' , nuclear physics b , volume 138 , issue 1 , p. 1 - 25 kentaro hori et.al . : `` mirror symmetry '' , ( clay mathematics monographs ; v.1 ) , american mathematical society and clay mathematics institute ( 2003 ) n.j . hitchin , g.b . segal and r.s . ward : `` integrable systems : twistors , loop groups , and riemann surfaces '' , ( oxford graduate texts in mathematics ; 4 ) , oxford ( 1999 ) anton kapustin : `` wilson-t hooft operators in four - dimensional gauge theories and s - duality '' , arxiv : hep - th/0501015 anton kapustin : `` chiral de rham complex and the half - twisted sigma - model '' , arxiv : hep - th/0504074 , anton kapustin , yi li : `` topological sigma - models with h - flux and twisted generalized complex manifolds '' , arxiv : hep - th/0407249 , sergei gukov , anton kapustin : `` new n=2 superconformal field theories from m / f theory orbifolds '' , nucl.phys . b545 ( 1999 ) 283 - 308 , anton kapustin , yi li : `` d - branes in topological minimal models : the landau - ginzburg approach '' , jhep 0407 ( 2004 ) 045 m. kontsevich , manin , yu : `` gromov - witten classes , quantum cohomology , and enumerative geometry '' , arxiv : hep - th/9402147 maxim kontsevich , lecture on motivic integration at orsay , 1995 , available at http://www.mabli.org/jet-bibliography.html anton kapustin , dmitri orlov : `` remarks on a - branes , mirror symmetry , and the fukaya category '' , arxiv : hep - th/0109098 masaki kashiwara , pierre schapira : `` sheaves on manifolds : with a short history @xmath342les debuts de la theorie des faisceaux@xmath343 by christian houzel '' , die grundlehren der mathematischen wissenschaften ; v. 292 ) , springer 1990 masaki kashiwara , pierre schapira : `` categories and sheaves '' , die grundlehren der mathematischen wissenschaften ; v. 332 ) , springer 2006 m. kapranov , e. vasserot : `` vertex algebras and the formal loop space '' , arxiv : math.ag/0107143 kapranov , m. ; vasserot : `` formal loops iv : chiral differential operators '' , arxiv : math.ag/0612371 anton kapustin , edward witten : `` electric - magnetic duality and the geometric langlands program '' , arxiv : hep - th/0604151 y. laszlo , `` hitchin s and wzw connections are the same '' , j. differential geom . 49 ( 1998 ) 547 - 576 gerard laumon : `` transformation de fourier generalisee '' , alg - geom/9603004 alastair craw : `` an introduction to motivic integration '' , arxiv : math.ag/9911179 , j. denef , f. loeser : `` motivic integration , quotient singularities and the mckay correspondence '' , arxiv : math.ag/9903187 fyodor malikov , vadim schechtman , arkady vaintrob : `` chiral de rham complex '' , arxiv : math/9803041 , commun.math.phys . 204 ( 1999 ) 439 - 473 nikita nekrasov : `` lectures on curved beta - gamma systems , pure spinors , and anomalies '' , arxiv : hep - th/0511008 edward witten : `` two - dimensional models with ( 0,2 ) supersymmetry : perturbative aspects '' , arxiv : hep - th/0504078 , vasily pestun , edward witten : `` the hitchin functionals and the topological b - model at one loop '' , arxiv : hep - th/0503083 edward witten : `` phases of @xmath48 theories in two dimensions '' , arxiv : hep - th/9603150	recently the chiral algebra of beilinson - drinfeld draws much attention in the mathematical physics of superstring theory . naively , this is a holomorphic conformal field theory with integer graded conformal dimension , whose target space not necessarily has the vanishing first chern class . this algebra has two ways of definition : one is that of malikov - schechtman - vaintrob by gluing affine patches , and the other is that of kapranov - vasserot by gluing the formal loop space . we will use the method of malikov - schechtman - vaintrob in order to compute the gerbes of chiral differential operators . in this paper , we will examine the two independent ansatzes of witten s ( 0,2 ) heterotic strings and nekrasov s generalized complex geometry are consistent in the case of @xmath0 , which has 3 affine patches and is expected to has the 1st pontrjagin anomaly . we also extend this direction to the case of 2 dimensional toric fano manifolds ( toric del pezzo surfaces ) of all degrees , by blowing up the generic 1,2,3 points of @xmath0 . these coincide with the computation of the hirzebruch riemann - roch theorem . the most notable case is the 1 point blowup , where the total gauge invariant anomaly vanishes . the significant future direction towards its application to the geometric langlands program is also discussed in the last section . beilinson - drinfeld chiral algebras for del pezzo surfaces 1.0 cm makoto sakurai + 0.5 cm _ department of physics , university of tokyo , 7 - 3 - 1 hongo , bunkyo - ku , tokyo 113 - 0033 , japan _
expressions are now derived for calculating the induced current within an arbitrary number of circular lenz lenses made from wires and placed within the incident magnetic field generated by an external rf coil . note that , by the theory of reciprocity @xcite , the expressions may be used in an equivalent sense to describe a situation in which a distinct volume of precessing magnetisation induces current within the lenz lenses , which in turn induce current within the external rf coil . that is , the lenz lenses and corresponding theory are applicable to both the excitation and receive chains of a conventional nmr ( or mri ) experiment . consider the simple arrangement of placing a single circular lenz lens coaxially at the midpoint between two elements of a helmholtz pair . faraday s law of induction states that the electromotive force ( emf ) , @xmath20 , generated in the lens is equal to the negative rate of change of magnetic flux @xmath21 that it encloses : @xmath22 lenz s law states that the sense of the current induced is such that it opposes this flux , hence the minus sign in equation ( [ faraday ] ) . in the limit that the gap size between the elements connecting the outer and inner loops of the lens is zero , the flux impressed on the lens is equal to that enclosed by the outer loop minus that enclosed by the inner loop : @xmath23 where @xmath24 is the current in the helmholtz pair , @xmath25 is the mutual inductance between the outer loop of the lens ( @xmath26 ) and one element of the helmholtz pair , and a similar definition holds for @xmath27 ( @xmath28 ) . we ignore capacitive effects and model the lenz lens itself as an @xmath29-circuit : @xmath30 where @xmath31 is the self - inductance of the lens ( with corresponding definitions for @xmath32 , @xmath33 and @xmath34 ) , @xmath35 is the resistance of the lens and @xmath36 is the induced current . equating the potential in equation ( [ voltage ] ) to the emf in equation ( [ faraday ] ) ( combined with equation ( [ phib ] ) ) and assuming a time - harmonic regime , we obtain the following expression for the induced current within the lens : @xmath37},\label{il1}\ ] ] where @xmath38 is the operating frequency ( radians / s ) of the rf coil . equation ( [ il1 ] ) represents a generalisation of an expression provided by schoenmaker et al . @xcite , which was derived for a lenz lens operating in the kilohertz regime . those authors ignored the resistance term in equation ( [ il1 ] ) , hence neglecting the frequency dependence of the induced current , and they also ignored the mutual inductance between the inner and outer loops of the lens . for high - frequency applications , such as the megahertz regime relevant to nmr , it is necessary to consider the skin - effect in the resistance and inductance calculations used in equation ( [ il1 ] ) . let us label the resistance of a wire under direct current to be @xmath39 and under ( high - frequency ) alternating current to be @xmath35 , as above . the ratio @xmath40 can be calculated exactly using kelvin - bessel functions @xcite , however this approach is computationally unstable when the skin - depth is small relative to the wire radius . a common alternative is to approximate the relative increase in resistance by the inverse of the relative decrease in the effective cross - sectional area of the ring defined by the skin - depth @xcite . this approach is accurate to within 5.5% of the exact result when the ratio of wire radius , @xmath41 , to skin - depth , @xmath42 , is greater than one @xcite , and below this limit the resistance is simply equal to @xmath39 : @xmath43 where @xmath44/(\pi a^{2})$ ] , in which @xmath45 is the radius of the outer loop of the lens and @xmath46 is the radius of the inner loop , @xmath47 , and @xmath48 and @xmath49 are the resistivity and permeability of the conductor , respectively ( e.g. for copper @xmath50 @xmath51 m and @xmath52 h / m ) . the self - inductance terms , @xmath33 and @xmath32 , may be calculated using the approximation @xcite : @xmath53,\label{l}\ ] ] where @xmath54 is the radius of the loop in question and the parameter @xmath55 depends on the frequency of operation . @xmath56 is the high - frequency limit for which current is restricted to the surface of the wire and @xmath57 is the low - frequency limit for which the current is uniform throughout the wire ; hence we set @xmath58 ( see equation ( [ rac ] ) ) . note that equation ( [ l ] ) is accurate to @xmath59 @xcite . the mutual inductances between the helmholtz coil and the elements of the lenz lens , and between the lens elements themselves , can be calculated using the formulae provided by babic et al . @xcite ( i.e. equations ( 24)-(25 ) ) . these formulae were derived following a vector potential argument and are applicable to any pair of arbitrarily placed circular conductors . note that matlab code for evaluating these formulae has been made available by babic et al . on their publisher s website ( http://ieeexplore.ieee.org ) . let us now consider the general case of multiple lenses placed arbitrarily within the vicinity of a helmholtz pair , as depicted in fig . [ fig : ll - overview](e ) . we must now treat the two loops of the helmholtz pair separately ( subscripts @xmath60 and @xmath61 ) and consider also the mutual inductances that exist between the elements of different lenses . after careful consideration of the enforced current sense between inner and outer loops of each lens , we arrive at the following expression for the flux contained by the @xmath62th lens ( @xmath63 ) : @xmath64 where , for example , @xmath65 is the mutual inductance between the outer loop of the @xmath62th lens and the inner loop of the @xmath66th lens , and @xmath67 is the induced current in the @xmath66th lens . similarly , the potential for the @xmath62th lens is given by : @xmath68i_{lk}.\label{vk}\ ] ] combining equations([faraday ] ) , ( [ phik ] ) and ( [ vk ] ) and rearranging , we obtain the following matrix equation : @xmath69 where @xmath70 and @xmath71 is a vector of length @xmath72 , which contains the induced current for each lens . note that it is straightforward to show that equations ( [ ilm])-([ilme ] ) reduce to equation ( [ il1 ] ) for the case of a single symmetrically placed lens . equations ( [ ilm])-([ilme ] ) permit the investigation of the dependence of the induced current(s ) and corresponding magnetic field on a variety of design parameters , such as the number of lenses , their geometry and the frequency of operation . as an illustrative example , equation ( [ ilm ] ) was solved for several different arrangements appropriate to the experiments described in section [ sec : expts ] . that is , we considered a helmholtz pair with a radius of 0.6 mm and a separation of 0.6 mm , carrying a current of 66 ma at 500 mhz . the wire for the lens(es ) was modelled using an effective circular cross - section with radius @xmath73 @xmath49 m . the radius of the outer loop of the lens(es ) was fixed at @xmath74 mm for all cases . the radius of the inner loop of the lens(es ) , @xmath46 , was set at either 0.1 mm or 0.2 mm and we considered both single lens and double lens cases , with a lens separation for the latter of either 0.1 mm or 0.2 mm . all simulations were performed using matlab^^ ( r2012b , mathworks^^ ) . in general , a smaller radius for the inner loop of a lens results in a stronger magnetic field at the midpoint . however , in nmr ( or mri ) it is desirable not only to have a strong transmit rf field ( and correspondingly high receive sensitivity ) but also to have a homogeneous field , such that the flip angle is constant throughout the region of interest ( roi ) . therefore the design of the lenz lens(es ) for nmr becomes an optimisation problem with the goal of maximising the induced field within the roi , with respect to the lens geometry , while maintaining an acceptable level of homogeneity . that is , we wish to solve : @xmath75 where @xmath76 is a vector containing the radii of the inner and outer loops and the axial positions of each lens , @xmath77 is the axial component of the total magnetic field calculated using the biot savart law , and @xmath20 is the specified acceptable average field error . as an illustrative example in keeping with the simulations outlined above , the roi was chosen to be a cylindrical volume of radius 0.06 mm and height 0.06 mm , centred at the origin and coaxial with the lenses and helmholtz coil . the mean terms in equation ( [ opt ] ) were evaluated by integrating numerically ( trapezoidal rule ) over the roi and then dividing by the volume . the optimisation in equation ( [ opt ] ) was achieved using the function ` fmincon ` from the matlab^^ optimisation toolbox(interior - point algorithm ) for three cases : one , two and four lenses . to improve convergence for the cases with multiple lenses , the optimisation was performed in two stages : firstly , the axial placements of the lenses were fixed at equal increments between the helmholtz coils and equation ( [ opt ] ) was solved with respect to the lens radii alone ; secondly , the results from the first optimisation were used as the starting guess for the full optimisation of equation ( [ opt ] ) . additional constraints were imposed to ensure that : the inner loops were always smaller than the outer loops ; the axial placements of multiple lenses were always separated by at least the wire diameter ; the axial placements of pairs of multiple lenses were always symmetric about @xmath78 and closer to the origin than the helmholtz coils ; the radii of the outer loops were always less than twice that of the helmholtz coil ( and all radii were positive ) . the optimisations took the order of seconds ( 1 lens , 7 s ; 2 lenses , 25 s ; 4 lenses , 106 s ) on a standard desktop computer ( 2.7 ghz intel^^ corei7 processor with 16 gb of ram ) . [ simfigs ] displays colour contour plots of the simulated @xmath79-component of the induced magnetic field over the plane @xmath80 for several different arrangements of lenz lenses within the helmholtz pair . similarly , fig . [ optfigs ] displays the results of solving the optimisation problem defined by equation ( [ opt ] ) for the cases of one , two and four lenses . -field induced by a helmholtz pair and a variety of different lenses : i.e. zero , one or two lenses of different inner loop size and/or axial position . the colour contour plots are all plotted on the @xmath81-plane as depicted in the illustrative example at the upper right of the figure . ] -plane by the blue dashed rectangle , whilst constraining the average field error to be 1% within this region of interest . the example with four lenses exhibits a threefold amplification of field sensitivity . ] symmetrical wire and plate - based lenz lenses according to figs . [ fig : ll - overview](b ) and [ fig : ll - overview](d ) with outer diameter ( od ) , and inner diameters ( ids ) of were patterned by means of photo - lithography and electroplating on glass substrates to be used with a custom micro helmholtz coil pair . the four designs fabricated are summarised in table [ tab : table1 ] , and short identifiers were assigned for easier discrimination of the different designs . .overview of the four different double lenz lens variants manufactured . [ cols="<,<,>,>,>",options="header " , ] [ tab : table1 ] to quantify the amplification of the lenses and to verify the theory developed above , we acquired a series of spin echo imaging experiments using a similar diameter micro helmholtz coil pair setup as presented in @xcite , but with one instead of two coil windings on each side . the lenz lens designs ll1-ll4 were primed with di - h@xmath82o and arranged in the setup illustrated in fig . [ fig : ll - overview](e ) before mri was performed at steadily decreasing attenuations . for reference , we further acquired mr images without a lenz lens present using comparable chips filled with h@xmath82o . a photograph of the actual setup and a brief summary of the results obtained is presented in supplementary fig . 1 . pulse parameters for the individual designs were determined by exporting the absolute signal values from paravision using a macro ( bruker ) . keeping the pulse length @xmath83 constant , the @xmath2-field required to generate a flip angle @xmath84 for a specific nucleus is given by @xmath85 which leads to @xmath86 for @xmath87 , and @xmath88 ( ) using the proton gyromagnetic ratio @xmath89 . however , due to the non - uniform field profile of the lens , a mixture of flip angles is generated . a global -pulse was defined by locating the highest signal amplitude along the sweeped power range , although the local flip angle might be at these particular pulse parameters . the probe efficiency @xmath90 , i.e. , the conversion efficiency of power into magnetic field for a certain probe ( lenz lens ) is given by @xmath91 and therefore @xmath92 according to equation ( [ eqn : snr ] ) . [ fig:90deg ] presents snr line profiles along the horizontal @xmath93-axis for ll1-ll4 and the reference scan without a lens . the snr was calculated by taking the ratios of the exported absolute values and the mean noise value obtained from a pixel matrix from the reference scan , as indicated in the figure . the line profiles were extracted from the images that were acquired using the individual , global -pulse parameters evaluated for each type . for ll1-ll4 the determined powers @xmath94 were , while for the reference scan , the -pulse was found at @xmath95 . these values lead to probe efficiencies @xmath96 of , for ll1-ll4 respectively , while the reference efficiency was @xmath97 . -axis , as exemplified by the broken line in the reference image . all scans were acquired at a global flip angle of . the mean noise was calculated from a pixel matrix in the upper right corner of the same scan , as depicted . statistics were derived from the regions illustrated by filled symbols . the powers for ll1 to ll4 and the reference scan were . ] note that the non - uniform field profile associated with each lens leads to a local sinusoidal modulation of the flip angle while sweeping the power . as such , while the global -pulse defined above results in the highest achievable signal amplitude , it does not lead to the highest achievable signal uniformity within the roi . the evolution of the snr profile of ll4 for @xmath98 is illustrated in fig . [ fig : ll4 ] , where @xmath99 was found at , and the profile clearly changes from convex to concave shape as the power is increased . the experimental results shown in figs . [ fig:90deg ] and [ fig : ll4 ] are summarised in table [ tab : table2 ] , which furthermore contains calculated figures of merit . data points defining the regions of interest , which were taken into account for calculations , are marked with filled symbols in both figures , while those values outside the rois are represented by hollow symbols . -1.69cm0 in @lrrrrrrrrrrrr design & @xmath3 & @xmath94 ( ) & @xmath84 & mean & max & std & @xmath100 & @xmath101 & @xmath102 & @xmath96 @xmath103 & @xmath104 @xmath103 & @xmath105 + ref . & 0 & 1.00 & & 17.2 & 18.4 & 0.7 & 18 & 1.0 & 1.0 & 0.53 & 0.53 & 1 + ll1 & 1 & 0.18 & & 47.7 & 52.4 & 3.2 & 5 & 2.4 & 2.8 & 1.25 & 1.48 & 1/36 + ll2 & 2 & 0.25 & & 30.2 & 36.2 & 3.4 & 12 & 2.0 & 1.8 & 1.06 & 0.95 & 1/9 + ll3 & 3 & 0.18 & & 42.7 & 44.8 & 1.5 & 5 & 2.4 & 2.5 & 1.25 & 1.33 & 1/36 + ll4 & 4 & 0.32 & & 27.5 & 32.3 & 2.8 & 12 & 1.8 & 1.6 & 0.94 & 0.85 & 1/9 + ll4 & 5 & 0.45 & & 24.5 & 29.3 & 1.2 & 12 & 1.5 & 1.4 & 0.94 & 0.74 & 1/9 + ll4 & 6 & 0.56 & & 23.2 & 26.8 & 3.2 & 12 & 1.3 & 1.3 & 0.94 & 0.69 & 1/9 + ll4 & 7 & 0.63 & & 15.0 & 17.6 & 2.4 & 12 & 1.3 & 0.9 & 0.94 & 0.48 & 1/9 + running index number . a global @xmath84 of was assumed at the maximum peak snr amplitude . deviating flip angles were calculated based on @xmath96 and by combining equations ( [ b1 ] ) and ( [ etap ] ) . number of voxels @xmath100 taken to calculate mean , max and std . the regions taken into account are represented by filled symbols in figs . [ fig:90deg ] and [ fig : ll4 ] . enhancement @xmath102 based on the ratio of the mean snr values with respect to the reference scan without any lens . effective probe efficiency @xmath106 with respect to the reference scan probe efficiency and the measured enhancements @xmath102 . ratio of the observable volumes of interest @xmath107 based on a constant slice thickness and on the various ids with respect to the reference scan without any lens . [ fig : fa ] and [ fig : spacing ] present comparisons between the experimental results and the corresponding simulations according to equations ( [ ilm])-([ilme ] ) . the simulated snr profiles were generated by first calculating the field amplification @xmath108 over an array of points : 31 points across the pixel in the @xmath109-direction , 65 points along the @xmath93-axis ( i.e. , one for each pixel ) , and 101 points across the slice in the @xmath79-direction . this field was averaged over the @xmath109- and @xmath79-directions , and the value ( @xmath110 ) either at the highest signal peak or at the centre - point in @xmath93 was used to define the flip angle . a corresponding snr array of points was then generated by applying @xmath111 . this snr array was subsequently averaged over the @xmath109- and @xmath79-directions to give the snr amplification factor along the @xmath93-axis . finally , the result was multiplied by the maximum snr of the reference scan , listed in the first row of table [ tab : table2 ] . [ fig : fa ] displays snr profiles for ll4 and demonstrates the impact of defining the global -pulse either at the location of the largest field amplification ( as per the experiment ) or at the centre - point ( cf . [ fig : ll4 ] ) . [ fig : spacing ] displays snr profiles for ll3 and demonstrates the impact of error in the spacing of the lenses ( i.e. versus ) . for both of the simulated profiles in fig . [ fig : spacing ] , the -pulse was defined at the centre - point , and for the case with spacing the array size used in the calculation was reduced to points . fig . [ simfigs ] demonstrates that the magnitude and homogeneity of the focused field is strongly dependent on the number and geometrical parameters of the lenz lenses . for a single lens , a smaller inner loop results in a stronger field at the cost of homogeneity . however , with the introduction of a second lens this homogeneity can be partially recovered , provided that an appropriate axial separation is chosen . for example , those pairs of lenses in fig . [ simfigs ] for which the ratio of the inner loop radius to the axial separation is equal to one ( i.e. akin to the helmholtz pair ; column 1 , row 3 and column 2 , row 4 ) clearly display a superior trade - off of amplification to homogeneity , over a spherical roi , compared to the other arrangements . however , to obtain the best trade - off over an arbitrarily shaped roi , especially when considering additional lenses , optimisation via equation ( [ opt ] ) is necessary . [ optfigs ] demonstrates that for an average relative field error of 1% within the roi defined in section [ ssec : meth_simopt ] , it is possible to achieve amplifications of 1.5 , 2.2 and 3.0 with one , two and four lenses , respectively , over the use of a helmholtz pair alone . if the field constraint is relaxed to 2% error , the amplification increases to 1.7 , 2.5 and 3.2 , respectively ( data not shown ) . notice that for the case of four lenses in fig . [ optfigs ] , the inner loops lie approximately on the surface of a sphere . however , for the equivalent case with 2% error , the two pairs of lenses are brought into close proximity to one another ( not shown ) . equations ( [ ilm])-([opt ] ) therefore allow the study and design of a wide variety of lens arrangements with flexible field constraints . indeed they may be applied to circular lenses of any orientation without necessarily being restricted to the centred coaxial arrangements considered herein . furthermore , the choice of considering a helmholtz pair as the transmit / receive rf coil has been made for demonstrative purposes only , since it was also used in some of the experiments , and the theory is applicable to any system for which the mutual inductance can be calculated . note that we have ignored capacitive effects that may be present for multiple closely - spaced lenses and also the finite gap between the elements that join the inner and outer loops . both of these factors will reduce the achievable amplification by some degree . note also that a direct method was used to solve equation ( [ opt ] ) , which was somewhat sensitive to the initial guess and hence the two - stage procedure described in section [ ssec : meth_simopt ] was implemented . an alternative approach would be to use stochastic optimisation , such as simulated annealing , to guarantee convergence to the global optimum at the cost of runtime . the experiments performed in this study clearly confirm the focusing effect and the associated local snr enhancement as predicted by the model . for the designs fabricated in this study , we achieved enhancements @xmath112 ( table [ tab : table2 ] ) . for the largest enhancement @xmath113 , the power required to generate a -pulse was reduced by a factor of compared to the reference . although in this case the observable volume of interest was reduced to of the initial volume , the gain in snr corresponds to a @xmath114-fold decrease in acquisition time , since @xmath115 @xcite . higher snrs were achieved for the plate - based designs due to the lower electrical resistance , i.e. , the increased noise contribution in the case of the wire - based design . mean snr enhancement may be estimated using the derived probe efficiencies @xmath96 , which are based on the -pulse power @xmath94 , and these were found to correspond well with the measured effective probe efficiencies @xmath104 ( table [ tab : table2 ] ) . a trade - off between the enhancement and the homogeneity of the snr profile was clearly apparent in the experimental results . for example , for ll4 at @xmath116 we observed an enhancement of @xmath117 and a standard deviation ( std ) of , whereas at @xmath118 the std improved by to at the cost of a decrease in the enhancement to @xmath119 ( see also fig . [ fig : ll4 ] ) . in the latter case , @xmath84 approached at the centre - point and exceeded at the peak field positions , which occurred close to the inner edge of the lenz lens defined by the inner diameter , and this resulted in a reduced snr at these points . the observed behaviour in the experiments is matched well by the theoretical model , as illustrated in fig . [ fig : fa ] for ll4 . for the case in which the -pulse corresponds to the peak field positions , the simulations are in high agreement with the results obtained in the experiment . furthermore , the simulated profile flattens out for the case in which the -pulse is at the centre - point , as expected from the experiments , and std decreases from . good agreement between simulations and experiment was also found for ll3 , as shown in fig . [ fig : spacing ] . the slightly higher snr values obtained in the experiment may be a result of fabrication tolerances , such as uncertainty in the spacing between lenses . in the present case , such tolerances may occur if the final thickness of the plated lenses is either not uniform or inaccurate , or if the thickness of the photoresist defining the spacing between the two lenses deviates from its nominal value . the example given in fig . [ fig : spacing ] shows that reducing the spacing from leads to a increase in mean snr . furthermore , the simulations assume line currents for the field calculation , whereas ll3 and ll4 are constructed using conductive material with a rectangular cross - section that has a low aspect ratio . therefore , the simulations may represent a lower bound to the trend observed in fig . [ fig:90deg ] from plate - based to wire - based lenses . to conclude , we have presented a novel method to locally increase the signal - to - noise ratio in nuclear magnetic resonance by using lenz lenses to focus the rf magnetic field of an nmr coil into a smaller volume of interest . as a result , higher spatial resolutions or reduced acquisition times become possible . the achievable enhancement strongly depends on the size ratios of the excitation coil and the dimensioning of the lens . the use of lenz lenses enabled not only the magnetic flux density to be intensified , but also the resulting field to be shaped by arranging multiple lenses in a pre - defined manner . the paper presented a theoretical description of a set of lenz lenses , which conformed closely to the experimental data obtained in a series of imaging experiments . the simulations enabled the optimisation of the design with respect to parameters such as signal amplitude and homogeneity . lenz lenses are useful not only for amplifying the @xmath2 magnetic field in the sample , but also for attenuating it in other regions , for example to avoid exciting regions with jumps in magnetic susceptibility , which would otherwise lead to distortions of the experiment , or to protect electronics from strong rf radiation . all simulations were performed using matlab^^ ( r2012b , mathworks^^ ) . the optimisation in equation ( [ opt ] ) was achieved using the function _ fmincon _ from the matlab^^ optimisation toolbox(interior - point algorithm ) for three cases : one , two and four lenses . the optimisations took the order of seconds ( 1 lens , 7 s ; 2 lenses , 25 s ; 4 lenses , 106 s ) on a standard desktop computer ( 2.7 ghz intel^^ corei7 processor with 16 gb of ram ) . [ fig : ll - fab ] illustrates the fabrication steps . at first , a cr / au seed layer was evaporated on two thick , 4-inch diameter float glass wafers ( d 263^^ t , schott ag , mainz , germany ) , where the cr - layer served as an adhesion promoter between the glass surface and the au - layer . subsequently , hexamethyldisilazane ( hmds ) was applied from the gas phase before spin - coating a layer of az9260 positive - tone photoresist ( microchemicals gmbh , germany ) . the wafers were stored in ambient atmosphere for around for rehydration , before windows for electroplating were opened using uv photo - lithography , which served as a mould structure . in the subsequent electroplating step , alignment marks and lenz lenses were structured accordingly by depositing a thick au - layer . in the case of the wire - type lenses , the width of the conductor was . an sem - closeup of an electroplated , wire - type lenz lens is presented in supplementary fig . 2 . afterwards , the mould layers were stripped using acetone before the au seed layer was etched using a potassium iodine / iodide based etchant . all etching and stripping steps were performed using megasonic agitation to ensure uniform wetting of the surface and homogeneous etching rates . before the subsequent uv - laser drilling step , dicing foil was laminated on both sides of one wafer to protect the substrate from debris during laser ablation . the latter was done using a uv - laser ( trumark 6330 , trumpf , germany ) , where 108 holes were drilled to realise two microfluidic ports for each of the 54 chips . after drilling , the fragile wafer was placed in a petri dish with isopropyl alcohol ( ipa ) , which dissolved the adhesive of the dicing foil , before the wafer was rinsed in deionized ( di ) water and dried using nitrogen gas . two layers of ordyl sy355 dry film resist ( elga europe s.r.l . , milano , italy ) were laminated onto the second substrate using a hot roll laminator ( mylam 12 , gmp , polch , germany ) . lamination was done at a speed of , a pressure of , and a temperature of approximately , resulting in a total nominal resist - thickness of . the resist was uv exposed at using a mask aligner ( ma6 , karl suss , germany ) to pattern microfluidic channels . after exposure , a post - exposure bake ( peb ) was performed for at . the structures were developed for about in bmr developer using ultrasonic agitation , followed by rinsing in ipa and di water . finally , the wafer was spin - dried . both substrates were subsequently bonded in a full wafer bonding process @xcite using a substrate bonder ( sb6 , karl sss , germany ) to realise the double lenz lens configuration , i.e. , parallel pairs of lenses . the wafers were aligned manually and fixed using the clamps of the sb6 chuck before loading it into the machine , where a tool pressure of was applied for at . the stack was hard baked for at before dicing chips with a nominal size of ( @xmath120 ) . the resulting spacing between the lenses of was hence defined by the thickness of the deposited au - layer and the height of the previously patterned ordyl photoresist layers . all magnetic resonance imaging experiments were performed on an avance iii nmr system ( bruker , rheinstetten , germany ) , controlled by the paravision^^ 5.1 imaging software ( bruker ) . the nmr scanner was operated at the proton larmor frequency of in combination with a micro 5 micro - imaging probe base and a micro 5 gradient system , driven at , which results in a maximum gradient strength of . throughout the experiments , the attenuation ( att ) was varied from ( @xmath121 ) in steps of . the relationship between att and @xmath18 is given by @xmath122 where @xmath95 and @xmath123 . acquisition parameters were set to : repetition time @xmath124 , echo time @xmath125 , flip angle @xmath99 , effective slice thickness @xmath126 , field of view @xmath127 , matrix @xmath128 and hence in - plane resolution , number of averages @xmath129 and an acquisition time ( ta ) per scan of . 10 a. g. webb . radiofrequency microcoils in magnetic resonance . , 31(1):142 , july 1997 . m. e. lacey , r. subramanian , d. l. olson , a. g. webb , and j. v. sweedler . high - resolution nmr spectroscopy of sample volumes from 1 nl to 10 l. , 99(10):31333152 , october 1999 . d. i. hoult and r. e. richards . the signal - to - noise ratio of the nuclear magnetic resonance experiment . , 24(1):7185 , october 1976 . t. l. peck , r. l. magin , and p. c. lauterbur . design and analysis of microcoils for nmr microscopy . , 108(2):114124 , august 1995 . h. d. w. hill and r. e. richards . limits of measurement in magnetic resonance . , 1(10):977 , october 1968 . d. l. olson , t. l. peck , a. g. webb , r. l. magin , and j. v. sweedler . high - resolution microcoil 1h - nmr for mass - limited , nanoliter - volume samples . , 270(5244):19671970 , december 1995 . j. schoenmaker , k. r. pirota , and j. c. teixeira . magnetic flux amplification by lenz lenses . , 84(8):085120 , august 2013 . d. i. hoult . . , 12(4):173187 , 2000 . d. i. hoult . . , 213(2):344346 , december 2011 . s. ramo , j. r. whinnery , and t. van duzer . . john wiley & sons , inc . , 3rd edition , 1994 . h. a. wheeler . formulas for the skin effect . , 30(9):412424 , 1942 . d. w. knight . practical continuous functions for the internal impedance of solid cylindrical conductors . http://www.g3ynh.info/zdocs/comps/zint.pdf ( version 2.06 . ) , apr 2013 . r. dengler . self inductance of a wire loop as a curve integral . arxiv:1204.1486v2 [ physics.class-ph ] , sep 2013 . s. babic , f. sirois , c. akyel , and c. girardi . mutual inductance calculation between circular filaments arbitrarily positioned in space : alternative to grover s formula . , 46(9):35913600 , 2010 . n. spengler , j. hfflin , a. moazenzadeh , d. mager , n. mackinnon , v. badilita , u. wallrabe , and j. g. korvink . heteronuclear micro - helmholtz coil facilitates m - range spatial and sub - hz spectral resolution nmr of nl - volume samples on customisable microfluidic chips . , 11(1):e0146384 , 01 2016 . a. macovski . noise in mri . , 36(3):494497 , 1996 . p. vulto , t. huesgen , b. albrecht , and g. a. urban . a full - wafer fabrication process for glass microfluidic chips with integrated electroplated electrodes by direct bonding of dry film resist . , 19(7):077001 , july 2009 . p. mller , n. spengler , h. zappe , and w. mnch . an optofluidic concept for a tunable micro - iris . , 19(6):14771484 , december 2010 . this work was partially funded by the european research council ( erc ) grant no . 290586 nmcel ( n.s . , m.v.m . , and j.g.k . ) and by an alexander von humboldt research fellowship ( p.t.w . ) . we further acknowledge michael pauls for recording the sem images , kay steffen for electroplating , and volker lehmann from bruker biospin for advice with the nmr instrumentation . n.s . and j.g.k . developed the concept of the study . n.s . , m.v.m . , and u.w . planned the microfabrication . n.s . and m.v.m . conducted the microfabrication , and performed the nmr experiments . p.t.w . derived the theory and performed the numerical optimisation experiments . all authors performed analysis and data interpretation , and drafted the manuscript . the authors declare no competing financial interests .	a high nmr detection sensitivity is indispensable when dealing with mass and volume - limited samples , or whenever a high spatial resolution is required . the use of miniaturised rf coils is a proven way to increase sensitivity , but may be impractical and is not applicable to every experimental situation . we present the use of magnetic lenses , denoted as lenz lenses due to their working principle , to focus the magnetic flux of a macroscopic rf coil into a smaller volume and thereby locally enhance the sensitivity of the nmr experiment at the expense of the total sensitive volume . besides focusing , such lenses facilitate re - guiding or re - shaping of magnetic fields much like optical lenses do with light beams . for the first time we experimentally demonstrate the use of lenz lenses in magnetic resonance and provide a compact mathematical description of the working principle . through simulations we show that optimal arrangements can be found . n.s . and p.t.w contributed equally to this work . ] both nuclear magnetic resonance ( nmr ) spectroscopy and magnetic resonance imaging ( mri ) suffer from an inherently low sensitivity . the signal strength is primarily determined by the equilibrium boltzmann distribution , with energy levels of the spin states just slightly above the thermal energy . consequently , the limit of detection ( lod ) is up to ten orders of magnitude worse compared to other analytical techniques @xcite . this fact severely limits the lowest detectable quantity in nmr spectroscopy and the highest achievable spatial resolution in mri , both being directly proportional to the signal - to - noise ratio ( snr ) of the experiment . the snr , first derived by hoult and richards @xcite , is given by @xcite @xmath0 where @xmath1 is a scaling factor to account for the homogeneity of the radio - frequency ( rf ) coil employed , @xmath2 is the rf - coil s magnetic field strength , @xmath3 the unit current , @xmath4 the coil s sensitivity , @xmath5 the observed sample volume , @xmath6 is the boltzmann constant , @xmath7 is the temperature , @xmath8 the spin density , @xmath9 the spin quantum number , @xmath10 is planck s constant @xmath11 divided by @xmath12 , @xmath13 is the larmor frequency determined by the gyromagnetic ratio @xmath14 of the nucleus of interest and the strength of the static magnetic field @xmath15 , @xmath16 is the electrical resistance of the coil contributing thermal noise , @xmath17 is the bandwidth of the receiver , and @xmath18 the power . the search for higher snr has resulted in steadily increasing @xmath15-field strengths , currently reaching a maximum of around ( @xmath19h larmor frequency ) for commercially available nmr systems , while field strengths of human scale mri magnets typically do not exceed in the clinic . while both the noise level and the boltzmann distribution would benefit from a reduced sample temperature , the freezing point and the operating temperature of the sample impose practical limits . non - equilibrium boltzmann polarisation factors can also be reached through spin order transfer techniques such as dynamic nuclear polarisation ( dnp ) or parahydrogen induced polarisation ( phip ) , which , however , often involve toxic substances and hence are not yet generally compatible with living biological samples . consequently , the snr is maximised by using optimised hardware along the receive path , such as dedicated rf receiver coils . such coil designs follow two strategies : ( i ) the filling factor @xcite and hence the magnetic interaction is maximised when the coil geometrically conforms to the sample as closely as possible , and ( ii ) the sensitivity increases linearly as the diameter of the coil decreases for a constant height - to - diameter ratio @xcite . the relative intrinsic snr is therefore higher when using a smaller , sample - adapted coil . hence such coils are used when recording mr images of body parts , e.g. , the brain , or when acquiring spectra from rare or volume - limited substances @xcite . however , additional equipment is required to cover the various cases , which is cost - intensive and requires maintenance , while situations remain where it is not yet possible to place the sample inside a dedicated active coil , such as when studying inner parts like organs . to circumvent these issues , we introduce here a method to locally improve the sensitivity of the nmr ( or mri ) experiment by means of broadband passive magnetic lenses . the working principle behind these lenses is geometry based and governed by lenz law , and hence they are referred to as lenz lenses @xcite . such lenses are capable of focusing the magnetic field of a macroscopic rf coil into a smaller spatial region , thereby locally increasing the flux density for the same applied radiofrequency current . lenz lenses can be simply made from wire , or conductive sheets such as copper foil , as shown in fig . [ fig : ll - overview ] , which illustrates two basic shapes . -2.5cm0 in from the outer edge to the inner edge , while reversing the flow direction , as further depicted in ( e ) , where two lenses are arranged in parallel in a helmholtz pair like configuration , denoted as a double lenz lens configuration.,title="fig : " ] due to their simple geometry and wireless , broadband inductive coupling , lenz lenses can easily be tailored to the intended application in a much more flexible manner than rf resonators , and are straightforward to fabricate . multiple lenses can be combined to enable the shaping or rotating of the magnetic field , similar to lenses in optics , and hence introduce a whole new degree of flexibility for the mr analysis of diverse samples .
spherically symmetric gravity coupled to a scalar field is a rich model , where one can test scenarios of black hole formation , the critical phenomena discovered by choptuik and hawking evaporation at the quantum level . for many years the full quantization of the model resisted analysis , in part due to the complicated nature of the hamiltonian structure of the system . initial attempts to study the problem were done by berger , chitre , nutku and moncrief @xcite and further developed by unruh @xcite . the resulting complicated nature of the gauge fixed hamiltonian led led unruh to say `` i present it here in the hope that someone else may be able to do something with it . '' more recently , husain and winkler and daghigh , kunstatter and gegenberg @xcite , using painlev gullstrand coordinates simplified somewhat unruh s treatment . none of these efforts provided a hamiltonian that was the spatial integral of a local density , leading to non - local equations of motion with the ensuing difficulty at the time of quantization . we recently noted that using ashtekar s new variables the construction of a local hamiltonian was possible . it was later suggested by unruh @xcite and gegenberg and kunstatter @xcite that a similar construction was possible in metric variables . in hindsight , this is not too surprising . the key element used in our construction was that in ashtekar s variables the gravitational part of the hamiltonian constraint becomes the total derivative of a quantity with respect to the radial variable . it turns out that some years ago , kucha @xcite introduced canonical coordinates for spherically symmetric vacuum gravity in which one of the coordinates is the mass as function of the radius . the gravitational part of the hamiltonian constraint in that case is given by the total derivative of the mass with respect to the radial variable . therefore a construction similar to the one we had carried out with ashtekar s variables can be carried out with kucha s variables . we will detail the construction here . as most gauge fixings , only certain families of initial data can be accommodated with a given choice of gauge . we set up a suitable initial - boundary value problem in the gauge fixed theory and for the physically important case of gaussian pulses . the organization of this paper is as follows . in section ii we discuss the gauge fixing in terms of kucha s variables . in section iii we set up the hamiltonian . in section iv we study the initial - boundary value problem , in particular for gaussian pulses . we end with a discussion of possibilities for quantization . the starting point is the three - metric in spherical coordinates , @xmath0 with @xmath1 and @xmath2 arbitrary functions of the radial variable ( and time ) , and their corresponding canonical momenta @xmath3 and @xmath4 . the canonical formulation in terms of these variables has been discussed by kucha @xcite , so we will not repeat it here , we refer the reader to his paper for details . the total hamiltonian density is obtained from the hamiltonian and diffeomorphism constraints , @xmath5 with @xmath6 the lapse and @xmath7 the shift . we now proceed to redefine the lapse and shift @xmath8 and from now on we drop the `` new '' subscripts . the total hamiltonian density can then be written with the gravitational part explicitly as a derivative with respect to the radial coordinate , @xmath9 + \frac{p_\lambda p_\phi \phi'}{\lambda r}\right\}+ n^r \left[\frac{1}{g}\left(-p_\lambda'\lambda+p_r r'\right)+p_\phi \phi'\right].\ ] ] let us now proceed to gauge fix . we start by setting @xmath10 . preserving this condition in time implies that the shift vanishes ( this is the rescaled shift , the original shift does not vanish ) . one solves the diffeomorphism constraint to obtain @xmath4 . to completely fix the gauge , we need to fix another variable . with that objective in mind , it is convenient to rewrite the hamiltonian as , @xmath11 with @xmath12 with @xmath13 at the moment just a constant , later it will be identified with the schwarzschild radius . the strategy for finding a gauge fixing that leads to a local hamiltonian will be to fix the value of the quantity @xmath14 . the resulting constraint therefore depends on the gravitational variables undifferentiated . when one preserves that gauge fixing in time , the lapse will be fixed by an algebraic equation rather than a differential one . this is the key point . if one were left as usual with a differential equation , the lapse would be an integral of the canonical variables . since the hamiltonian is an integral that involves the lapse , it becomes an integral of an integral and in that sense is non - local . so we choose @xmath15 . preservation in time of this condition determines the lapse as an algebraic function of @xmath16 . we proceed to solve the variable @xmath1 through the gauge fixing , @xmath17 with @xmath18 and substitute it in the total hamiltonian , which leads to , @xmath19 which we should solve to get @xmath3 as a function of @xmath20 and @xmath21 . we will see later how to do this in a compact way . this completes the gauge fixing . the free variables are @xmath22 . we now go to the evolution equations for those variables , derived before the gauge fixing , and substitute the latter . the resulting equations can be shown to be equivalent to those that stem from a true hamiltonian , @xmath23 with @xmath24 and in these expressions @xmath3 should be substituted by the expressions we derived before during the gauge fixing . a constructive procedure to directly obtain the above true hamiltonian is to perform a canonical transformation from the variables @xmath25 to a new set of variables @xmath26 . this should be done before the gauge fixing @xmath15 , so at the moment @xmath14 is function of @xmath1 and @xmath3 given by ( [ 9 ] ) . this motivates us to consider a generating function of type 3 , @xmath27 for which one would have that , @xmath28 and solving for @xmath1 in the definition of @xmath14 ( [ 9 ] ) this can be integrated to give @xmath29 with @xmath30 being the non - gauge fixed version of @xmath31 , @xmath32 we therefore have for the conjugate variable , @xmath33 and for @xmath1 , @xmath34 the total hamiltonian in terms of the new variables is @xmath35 with @xmath3 obtained by solving ( [ 8 ] ) with @xmath1 substituted by ( [ 17 ] ) . we now proceed to gauge fix @xmath15 . preservation in time of this condition leads to @xmath36 . noting that @xmath22 have vanishing poisson brackets with @xmath37 , if we write the evolution equations and substitute the gauge fixing in them , we have that , @xmath38 with @xmath39 as in any gauge fixing in a complicated theory like general relativity , one does not expect one will cover all of phase space . the limitation here is given by the equation for @xmath3 , which written explicitly reads , @xmath40 with @xmath41 this will not generically yield a real value for @xmath3 given arbitrary initial data for @xmath22 . this not only limits the initial data but also the boundary conditions one can give at the outer and inner boundary . so from now on we are limited to consider more specific situations . one has certain freedom to modify things by playing with the function @xmath42 that determines the gauge fixing . for different choices of @xmath42 different families of initial and boundary data will be acceptable as producing real values for the variables . a case of great interest is the study of the propagation of wave packets of scalar field on a black hole space - time . we will therefore concentrate ourselves on that situation . this will require specifying at spatial infinity boundary conditions such that the geometry is asymptotically that of schwarzschild with no ingoing matter fields , and the inner boundary corresponds to a dynamical horizon with matter fields purely ingoing into it . we choose as initial data for the scalar field , @xmath43 where we are considering a gaussian pulse and we added a factor @xmath44 such that the field vanishes at the horizon initially . this makes the horizon for the initial data an isolated one . for @xmath21 we choose what is needed to have a purely ingoing pulse , @xmath45 we will now proceed to fix the gauge in such a way that the bi - quadratic equation ( [ 24 ] ) has at least a pair of real roots . notice that in ( [ 24 ] ) both @xmath46 and @xmath47 are always positive . for the quadratic equation for @xmath48 have a positive root one needs to make the linear term negative . one possible strategy is to consider ( [ 28 ] ) and integrate it using the initial data we are considering , @xmath49 the right hand side is a bit complicated , but evaluating it numerically one sees it has the form of a gaussian - like shape . one can therefore simply take for @xmath50 a gaussian that envelops the integral as a gauge choice , @xmath51 with @xmath52 appropriately big for it to envelop the integrand . the integral for @xmath50 can be evaluated in closed form , but its expression is lengthy . the form of the function is relatively simple , it is a modified step function as shown in the figure . , @xmath53 , @xmath54 , @xmath55.,height=226 ] in terms of @xmath50 one can solve ( [ 24 ] ) for @xmath3 . the closed form expression is again lengthy . asymptotically for large @xmath56 we have @xmath57 with @xmath58 a constant . this completes the determination of the initial data . we need to fix the gauge for all time . to do this , we consider the preservation of @xmath59 , which determines the lapse . we would like the lapse , at least asymptotically , to reproduce the usual manifestly asymptotically flat nature of the schwarzschild space - time . this corresponds to @xmath60 ( recall that we are referring to @xmath61 , which corresponds asymptotically to @xmath62 ) . this results , asymptotically in @xmath63 , where @xmath64 is the asymptotic value of the expression of @xmath50 , which , for instance , can be read off for large values of @xmath56 in the figure above . the expression of @xmath65 is , @xmath66 with @xmath67 a constant . with this form of the gauge fixing , the asymptotic form of the metric is , @xmath68 with @xmath69 , which shows that the asymptotic mass is the same as that of the horizon plus the contribution of the scalar field . with a simple redefinition of @xmath70 this yields the usual expression of the schwarzschild metric in the schwarzschild coordinates . although we have not studied the evolution in detail , one can envision using a gauge with ( [ 32 ] ) modified to be an ingoing pulse and this should yield real expressions for all quantities as the pulse travels inward , at least far away from the black hole . we have shown that one can gauge fix spherically symmetric gravity coupled to a scalar field in terms of the traditional metric variables with a hamiltonian that is the integral of a local density in explicit form . we construct a family of gauge fixings that can accommodate ingoing gaussian pulses and show that they include manifestly asymptotically flat coordinates . the construction of the gauge is such that it is clear that it will evolve correctly in time , at least for a limited amount of time . it should be noted that we have not analyzed properly the inner boundary condition beyond the initial slice . one presumably would like to have a dynamical horizon that increases its mass as the ingoing pulses progress towards the black hole , at least studying the problem classically . quantum mechanically , it is less clear what one needs at the horizon , since hawking radiation should be present . the question of quantization of the model in terms of this gauge implies having to deal with the quartic equation ( [ 24 ] ) that generically will lead to complex values . it is therefore unclear that the constructed hamiltonian could be promoted to a self - adjoint operator . it should be noted that there exist techniques @xcite to deal with these types of issues in quantization . we have recently illustrated this in a model system @xcite . they are however , limited to certain regimes . realistically , this type of approach is unlikely to yield insights about extreme regimes like the ones close to the singularity . but it may be useful in other situations , like in those in which a large black hole emits hawking radiation to study , for instance , the back reaction of the weak radiation on the large black hole . this work was supported in part by grant nsf - phy-0968871 , funds of the hearne institute for theoretical physics , cct - lsu , pedeciba and anii pdt63/076 . this publication was made possible through the support of a grant from the john templeton foundation . the opinions expressed in this publication are those of the author(s ) and do not necessarily reflect the views of the john templeton foundation . k. giesel and t. thiemann , class . * 27 * , 175009 ( 2010 ) [ arxiv:0711.0119 [ gr - qc ] ] ; r. gambini , j. pullin , in preparation . r. gambini , j. pullin , `` self - adjointness in the hamiltonians of deparameterized totally constrained theories : a model '' [ arxiv:1207.5730 [ gr - qc ] ]	we discuss a gauge fixing of gravity coupled to a scalar field in spherical symmetry such that the hamiltonian is an integral over space of a local density . in a previous paper we had presented it using ashtekar s new variables . here we study it in metric variables . we specify completely the initial - boundary value problem for ingoing gaussian pulses .
it is the characteristic feature of soft matter systems that a macromolecular component of nano- to micrometer size is dispersed in a solvent of much smaller molecules . the mesoscopic length scale of the dispersed component implies that crystalline phases have a very small shear modulus which roughly scales like the inverse of the third power of the structural length scale and that both crystalline and fluid phases are characterized by long structural relaxation times . soft matter systems have therefore interesting dynamical properties , because the time scale of an external perturbation can easily become comparable with the intrinsic relaxation time of the dispersed macromolecules . one of the unique properties of soft matter is its viscoelastic behavior@xcite . due to the long structural relaxation time , the internal degrees of freedom can not relax sufficiently fast in an oscillatory shear flow , so that there is some elastic restoring force which pushes the system back to its previous state . a very well studied example of viscoelastic fluids are polymer solutions and polymer melts@xcite . in the case of polymer melts , the characteristic time scale is given by the reptation time , _ i.e. _ by the time it takes a chain to slide by its contour length along the tube formed by other polymer chains@xcite . in order to bridge the length- and time - scale gap between the solvent and macromolecular or colloidal scales , several mesoscopic simulation techniques such as the lattice - boltzmann method , dissipative - particle dynamics ( dpd ) , and multi - particle collision dynamics ( mpc ) have been suggested in recent years , and are in the process of being developed further . the idea of all these methods is to strongly simplify the microscopic dynamics in order to gain computational efficiency , but at the same time to exactly satisfy the conservation laws of mass , momentum and energy , so that hydrodynamic behavior emerges naturally on larger length scales . we will focus here on the multi - particle collision dynamics ( mpc ) technique@xcite , also called stochastic rotation dynamics@xcite ( srd ) , originally developed for newtonian fluids . this particle - based hydrodynamics method consists of alternating streaming and collision steps . in the streaming step , point particles move ballistically . in the collision step , particles are sorted into the cells of a simple cubic ( or square ) lattice . all particles in a cell collide by a rotation of their velocities relative to the center - of - mass velocity around a random axis@xcite . a random shift of the cell lattice is performed before each collision step in order to restore galilean invariance@xcite . this method has been applied very successfully to study the hydrodynamic behavior of many complex fluids , such as polymer solutions in equilibrium@xcite and flow@xcite , colloidal dispersions@xcite , vesicle suspensions@xcite , and reactive fluids@xcite . the viscoelastic behavior of polymer solutions leads to many unusual flow phenomena , such as shear - induced phase separation@xcite , viscoelastic phase separation@xcite , and elastic turbulence@xcite . a coarse - grained description of viscoelastic fluids is necessary in order to obtain a detailed understanding of the role of elastic forces in such flow instabilities . however , there is a second level of complexity in soft matter system , in which a colloidal component is dispersed in a solvent , which is itself a complex fluid . examples are spherical or rod - like colloids dispersed in polymer solutions or melts , which are exposed to a shear flow@xcite . shear flow can induce particle aggregation and alignment in these systems . this is important , for example , in the processing of nanocomposites@xcite . the aim of this paper is therefore the development of a mpc algorithm , which is able to describe viscoelastic phenomena , but at the same time retains the computational simplicity of standard mpc for newtonian fluids , and thereby allows to take advantage of this mesoscale simulation for the investigation of flow instabilities as well as suspensions with viscoelastic solvents . we show that this goal can be achieved by replacing the point particles of standard mpc by harmonic dumbbells . in order to obtain a strong elastic contribution , we consider a fluid , which consists of dumbbells only . however , it is of course straightforward to mix dumbbells with a point - particle solvent . a similar idea has been suggested recently for dpd fluids@xcite . in our mpc model , we consider @xmath0 point particles of mass @xmath1 , which are pairwise connected by a harmonic potential @xmath2 to form dumbbells , where @xmath3 is the spring constant . the center - of - mass position @xmath4 and velocity @xmath5 for each dumbbell @xmath6 , with @xmath7 , are represented by @xmath8 here @xmath9 , @xmath10 and @xmath11 , @xmath12 denote the position and velocity of the two point particles composing a dumbbell @xmath6 , respectively . the mpc algorithm consists of two steps , streaming and collisions@xcite . in the streaming step , within a time interval @xmath13 , the motion of all dumbbells is governed by newton s equations of motion , @xmath14 where @xmath15 is the mass of a dumbbell , and @xmath16 is the total external force on dumbbell @xmath6 . we consider only constant force fields . the center - of - mass positions and velocities of dumbbells are then given by a simple ballistic motion . the evolution of the relative coordinates of each dumbbell are determined by the harmonic interaction potential , so that @xmath17 with angular frequency @xmath18 . the vectors @xmath19 and @xmath20 are different for each time step , and are calculated from the relative positions and velocities of the point particles of dumbbell @xmath6 before the streaming step , @xmath21 in the mpc algorithm described here , @xmath22 , @xmath23 , @xmath24 and @xmath25 are continuous variables , evolving in discrete increments of time . in the absence of shear flow , the average length of the dumbbell is @xmath26 for a @xmath27-dimensional system . in the collision step , the point particles are sorted into the cells of a cubic lattice with lattice constant @xmath28 . multi - particle collisions are performed for all particles in a cell @xmath29 , by the same srd algorithm@xcite as for point particle fluids . the velocity of each particle relative to the center - of - mass velocity @xmath30 of the cell is rotated around a randomly chosen axis by a fixed angle @xmath31 , @xmath32\;,\ ] ] where @xmath33 is a stochastic rotation matrix , and @xmath34 with @xmath35 the number of particles within cell @xmath29 . this step guarantees that each particle changes the direction as well as the magnitude of its velocity during the multi - particle collisions , while the local momentum and the kinetic energy are conserved . random shifts are applied in each direction , so that the galilean invariance is ensured even in case of small mean free path@xcite . in order to describe couette or oscillatory shear flow , the system is confined within two parallel hard walls in the @xmath36 direction , which are moving oppositely along the @xmath37 direction . here , @xmath38 , @xmath39 and @xmath40 are used to denote the dimension of the simulation box along the corresponding directions . for a steady shear flow , the shear rate is given by @xmath41 , with @xmath42 the @xmath37 component of the velocity of the wall moving along the positive direction . periodic boundary conditions are applied in the @xmath37 and @xmath43 direction , bounce - back boundary condition in the @xmath36 direction . the system is therefore divided into @xmath44 and @xmath45 cells in the @xmath37 and @xmath43 directions ( parallel to the walls ) , but @xmath46 cells in the @xmath36 direction because of the random shifts . at the walls , for collision cells which are not completely filled by particles , extra virtual point particles are added to conserve the monomer number density , @xmath47 , defined by the average number of monomers per cell@xcite . in principle , the velocities of the virtual particles can be drawn from a maxwell - boltzmann distribution of average velocity equal to the wall velocity and variance @xmath48 , where @xmath49 the bulk temperature . in the simulation code , it is not necessary to sample the velocity of virtual wall particles individually . a random vector from maxwell - boltzmann distribution with wall velocity and variance @xmath50 is then used instead of the contribution of the entire virtual particles in the cell , where @xmath51 is the number of real particles in that cell . for point particles , the combination of bounce - back boundary condition and virtual wall particles has been shown to guarantee no - slip boundary condition to a very good approximation@xcite . in order to keep the system temperature constant , various thermostats can be employed . in the first case , the mpc method with collisions by stochastic rotations ( mpc - srd ) of relative velocities is augmented by velocity rescaling . the simulation box is subdivided into @xmath52 layers parallel to the walls . in each layer , the new velocity @xmath53 of each particle @xmath54 in cell @xmath29 is obtained by rescaling the velocity relative to the center - of - mass velocity of that cell , @xmath55 here @xmath56 is calculated from the actual velocity distribution @xmath57 where @xmath35 denotes the number of particles in cell @xmath29 and @xmath58 the number of cells which contains particles within a layer . in the second case , the anderson s thermostat version of mpc , denoted mpc - at , is applied@xcite . this thermostat employs a different collision rule instead of eq . ( [ vcm ] ) . in the mpc - at@xmath59 version of the algorithm ( without angular momentum conservation , compare sec . [ sec : angular ] below ) , the new velocities of point particles in the collision step are assigned as@xcite @xmath60 here @xmath61 is a velocity chosen from the maxwell - boltzmann distribution and @xmath62 the number of particles within cell @xmath63 . instead of energy conservation in mpc , the temperature is kept constant in mpc - at . the standard mpc algorithm as well as the anderson thermostat version do not conserve angular momentum . it has been shown recently@xcite that this lack of angular - momentum conservation may lead to quantitative or even qualitative incorrect results , like non - physical torques in circular couette flows . we therefore also consider the angular - momentum conserving modification of mpc - at@xcite , denoted mpc - at@xmath64 . here , the velocities in the collision step are calculated by @xmath65 where @xmath66 and @xmath67 denote the moment - of - inertia tensor and the center of mass of particles in the cell , respectively . in the absence of shear flow , the monomer density profile @xmath68 can be calculated from the interaction potentials @xmath69 of the dumbbells , @xmath70 \;,\end{aligned}\ ] ] where @xmath71 is the bulk monomer density , @xmath72 the partition function , and _ erf _ the error function . fig . [ den ] shows excellent agreement of the theoretical prediction ( [ wall0 ] ) with simulation data . the particles are not equally distributed along the wall direction ; instead , at both walls , the density is only half of the bulk density . in order to reduce possible slip effects , it seems desirable to make the particle distribution as uniform as possible . an attractive potential is therefore applied when the center - of - mass position of the dumbbells approaches one of the walls , @xmath73 where @xmath74 is the one - dimensional average extension of a dumbbell . the density profile is now given by @xmath75 the advantages of the piecewise linear form ( [ wall1 ] ) of the wall potential are twofold . firstly and most importantly , it allows for an analytical integration of the equations of motion during the streaming step . secondly , the density profile in the absence of flow can again be calculated analytically ( see appendix for details ) . ( color online ) monomer density profiles with ( squares ) and without ( circles ) attractive wall potentials applied along the wall direction when particles approach close to walls . the dashed and dotted lines are the theoretical prediction described in eq . ( [ wall ] ) and eq . ( [ wall2 ] ) , respectively . the spring constant of dumbbells and the collision time are @xmath76 and @xmath77 , respectively . both simulations are with absence of shear flow . ] ( color online ) monomer density profiles at various dimensionless shear rates @xmath78 , ranging from @xmath79 to @xmath80 , when attractive wall potentials are applied . the spring constant of dumbbells and the collision time are as same as those in fig . [ den ] . the small ripples in the profile at large @xmath81 are due to inhomogeneities in the temperature profile , since velocity rescaling is not sufficiently efficient at high shear rates . the ripples to not appear for mpc - at . ] the simulated density profile shows excellent agreement with the analytical solution of eqs . ( [ wall1 ] ) and ( [ wall2 ] ) ( see appendix ) . the factors @xmath82 and @xmath83 are chosen to obtain a nearly uniform density distribution . this is achieved for @xmath84 and @xmath85 . as shown in fig . [ den ] , the densities of point particles at both wall boundaries deviate by less than @xmath86 lower from the bulk value , when the attractive wall potentials are applied . simulations are also performed on systems of dumbbells with various spring constants , ranging from @xmath87 to @xmath88 , in the absence of shear flow . it is found that for the given values of @xmath82 and @xmath83 , the density profiles are essentially independent of the spring constant of the dumbbell in the range @xmath89 . in fig . [ rho ] , we plot density profiles in shear flow . at lower shear rates , _ i.e. _ @xmath90 , nearly identical profiles are obtained as without flow . for higher shear rates , deviations of the density profile from the equilibrium profile become significant . nevertheless , these profiles are still more uniform than those without an attractive wall potential . our investigations are mainly focusing on relatively low shear rates , where the non - uniformity of the density profile is not significant . in the mpc model , the viscosity @xmath91 consists of a kinetic and collisional contribution@xcite . at steady shear rates , with flow along the @xmath37 direction and gradient along the @xmath36 direction , @xmath91 is calculated by measuring the @xmath92 component of the stress tensor , @xmath93 , so that @xmath94 . in the streaming step , @xmath95 is proportional to the flux of the @xmath37 momentum crossing a plane normal to the @xmath36 direction . since the stress tensor is independent of the position of the plane , we choose @xmath96 or @xmath97 to measure the momentum transfer . in two - dimensional simulations , @xmath98\;,\ ] ] where @xmath99 $ ] is the time at which particle @xmath6 bounces back from the wall , @xmath100 and @xmath101 are the velocities just before and after the collision with the wall , and @xmath102 denotes the number of particles which hit one of the walls in the time interval @xmath103 $ ] . in the collision step , particles close to the wall will change their velocities due to the multi - particle collisions with virtual wall particles with average velocity @xmath104 , so that @xmath105\;.\ ] ] here @xmath106 denotes the number of particles which have multi - particle collisions with virtual particles , while @xmath107 and @xmath108 are the velocities of particle @xmath6 before and after the collision step , respectively . in our simulations , @xmath106 is found to be much larger than @xmath102 for small collision times @xmath13 , indicating that the collisional part dominates the shear viscosity . simulations are first performed on a system of pure point - like fluid particles to verify the measurement of the zero - shear viscosity from eqs . ( [ stresskin ] ) and ( [ stresscol ] ) . we get perfect agreement with the theoretical predictions@xcite for @xmath91 . the shear viscosity can also be measured from system under poiseuille flow@xcite by @xmath109 where @xmath110 is the gravitation field , and @xmath111 the maximum flow velocity . in an oscillatory shear flow , the shear rate @xmath112 is time - dependent , @xmath113 where @xmath114 and @xmath115 are the strain amplitude and the oscillation frequency , respectively . note that the frequency @xmath115 in eq . ( [ dgt ] ) is independent of the angular frequency @xmath116 of harmonic dumbbells in section [ sec : model ] . in our simulations , we choose @xmath117 in order to investigate the linear viscoelastic regime . the stress tensor is divided into two contributions , the viscous part @xmath118 and the elastic part @xmath119 , so that@xcite @xmath120\;,\end{aligned}\ ] ] where @xmath121 is the storage modulus , which measures the in - phase storage of the elastic energy , and @xmath122 is the loss modulus , which measures the out - of - phase energy dissipation . for a simple maxwell fluid , @xmath121 and @xmath122 are given by@xcite @xmath123 where @xmath124 is a characteristic relaxation frequency , and @xmath125 is a characteristic shear modulus . in the limit of @xmath126 , the loss modulus is @xmath127 , where @xmath91 is the zero - shear viscosity . in order to estimate the rheological properties of our model fluid , we modify the kinetic theory for dilute solutions of elastic dumbbells @xcite . for hookean dumbbells in a solvent , the viscosity @xmath128 , the storage modulus @xmath129 and the loss modulus @xmath130 are given by@xcite @xmath131 @xmath132 @xmath133 where @xmath134 with solvent viscosity @xmath135 and friction coefficient @xmath136 of a monomer . moreover , the expectation value for the square of the monomer separation , divided by its equilibrium value , is given by@xcite @xmath137 in ref . , the friction coefficient is obtained from stokes law for a bead of radius @xmath138 in the solvent , _ i.e. _ @xmath139 . however , in the mpc dumbbell fluid , there exists no explicit solvent and the monomers are point particles instead of spheres . nevertheless , the motion of the monomers is governed by the friction caused by the surrounding monomers which can be considered to take the role of the solvent . using @xmath140 , which follows from the stokes - einstein relation , we can thus relate the friction to the diffusion constant @xmath141 of a mpc fluid of point particles with the same monomer density . similarly , we substitute the viscosity of the solvent , @xmath135 , by the corresponding viscosity @xmath142 of a mpc fluid of point particles . theoretical expression for @xmath142 and @xmath141 for the different collision methods can be found in refs . and refs . , respectively . the zero - shear viscosity then reads @xmath143 where we have introduced @xmath144 note that the limit @xmath145 corresponds to a mpc fluid of @xmath146 point particles of mass @xmath147 . here , the second term in eq . ( [ eq : visc_theo_mpc ] ) vanishes , and since @xmath148 for not too small @xmath47 and sufficiently small @xmath13 ( so that the collisional part of the viscosity dominates ) , the viscosity resulting from this simple theory approaches the correct value in this limit . consequently , we use the same substitutions for the storage and loss modulus , and for the average dumbbell extension , and obtain @xmath149 @xmath150 and @xmath151 we emphasize that the above expressions only serve as a semi - quantitative description of the mpc dumbbell fluid . for example , the employed expressions for the diffusion constant neglect hydrodynamic interactions , which become important for small time steps @xmath13 . in the remainder of this article , we introduce dimensionless quantities by measuring length in unit of the lattice constant @xmath28 , mass in unit of the dumbbell mass @xmath152 , time in units of @xmath153 , velocity in units of @xmath154 , monomer number density @xmath47 in units of @xmath155 , where @xmath27 is the spatial dimension , and the spring constant @xmath63 in units of @xmath156 . the shear rate @xmath157 and all kinds of frequencies are measured in units of @xmath158 . finally the viscosity @xmath91 is in units of @xmath159 , and the storage modulus @xmath121 and the loss modulus @xmath122 are in units of @xmath160 . in these dimensionless units , the mean free path @xmath161 ( in units of the lattice constant ) becomes equivalent to the time step @xmath13 . in our simulations , harmonic dumbbells with @xmath162 point particles are initially placed in a two- or three - dimensional rectangular box at random . we choose the average number density of point particles @xmath163 and @xmath164 for all two - dimensional simulations which results in @xmath165 . the collision time ranges from @xmath166 to @xmath167 , while the spring constant ranges from @xmath168 to @xmath169 . the rotational angle is chosen @xmath170 and @xmath171 for two- or three - dimensional simulations , respectively . we use small @xmath13 and large @xmath31 to obtain large schmidt numbers required for fluid - like behavior@xcite . most of the results shown are obtained from two - dimensional systems , except in a few cases where this is explicitly mentioned . in tab . [ tab : omegah ] , the theoretical values for the diffusion constant @xmath141 are given for @xmath172 for the different collision methods and various monomer densities @xcite . the corresponding results for other time steps @xmath13 can be obtained by employing the linear relationship between @xmath141 and @xmath13 . .diffusion constants @xmath141 of point - particle fluids for the standard mpc - srd algorithm , as well as for mpc - at@xmath59 and mpc - at@xmath64 simulations for various monomer densities , in two dimensions . all data are calculated for collision time @xmath172 . diffusion constants for other time steps @xmath13 can be obtained by employing the linear relationship between @xmath141 and @xmath13 . note that the values for mpc - at@xmath59 are identical with those for mpc - srd with collision angle @xmath170 . [ cols="^,^,^,^",options="header " , ] ( color online ) snapshots of dumbbell configurations in steady shear flow . the system size is @xmath173 . half of each dumbbell is colored red , the other half yellow for reason of visualization . in each frame only 2500 dumbbells are shown , so that the density is @xmath174 times as high as appears from the pictures . the spring constant and the collision time are @xmath76 and @xmath77 , respectively . from ( a ) to ( d ) , the applied shear rates are @xmath175 , @xmath176 , @xmath177 and @xmath178 . , width=264 ] in fig . [ snap ] , we present snapshots for steady shear flow with a simulation box containing 25000 dumbbells . at lower shear rates , _ i.e. _ @xmath179 , see fig . [ snap]a and [ snap]b , the average extension of the dumbbells is hardly distinguishable from the equilibrium value . in these two cases , the shear flow is not strong enough to align the dumbbells along the flow direction , so that both systems are still isotropic . with increasing @xmath157 , shear forces overwhelm entropic forces . as a result , dumbbells are largely stretched , at the same time reorientated along the flow direction , as presented in fig . [ snap]c and [ snap]d . note that near both the walls , the average size @xmath180 of the dumbbells in flow is larger than in the bulk . also , an alignment of the dumbbells is found near the walls , both with and without shear flow , with peaks at @xmath96 and @xmath97 . this is an effect of the geometrical constraints imposed on anisotropic particles by a hard wall . furthermore , a maximum of the extension occurs at a _ finite _ distance from the wall , which we attribute to the combined effect of the wall and the flow conditions ; dumbbells very close to the wall are sterically oriented parallel to the wall and thus experience only a very small shear force , while those a little further away are close to the average inclination angle ( see fig . [ e2e ] below ) , which corresponds to the largest stretching . the distance of the position of the maximum from the wall decreases with increasing shear rate , and seems to approach the size of the collision cells for large @xmath181 . the relative peak height increases with increasing shear rate . for example , we find that the maximum extension @xmath180 near the wall is about @xmath182 larger than the bulk extension for @xmath183 , while it is about @xmath184 larger than in the bulk for @xmath185 . [ e2e ] . each dot indicates the end - to - end vector of a dumbbell . , title="fig:",width=264 ] + fig . [ e2e ] presents the extensional and orientational distribution of the dumbbells for various shear rates . at lower shear rates , @xmath186 , the end - to - end vector of the dumbbells is distributed on a circle , see fig . [ e2e]a and b , indicating an isotropic orientation . at a higher shear rate , @xmath187 , the orientational distribution becomes an elongated ellipse , see fig . with increasing shear rate , the distribution elongates further . simultaneously , dumbbells become more aligned with the flow direction , as can be seen quantitatively from the inclination angle @xmath188 shown in fig . [ angle ] . here , the inclination angle is defined as the angle between the average orientation of the end - to - end vector of a dumbbell and the the flow direction . at lower shear rates , @xmath186 , the inclination angle approaches @xmath189 , while it decays to zero for large shear rates with a power law @xmath190 . ( color online ) the inclination angle @xmath188 as a function of dimensionless shear rate @xmath191 for the system of fig . [ snap ] , with spring constant @xmath76 , collision time @xmath77 , and system size @xmath173 . ] ( color online ) shear viscosity @xmath91 and scaled average dumbbell length @xmath192 as a function of dimensionless shear rate @xmath191 . systems with the wall separation @xmath193 , and @xmath194 are investigated . the spring constant is @xmath76 and the collision time @xmath77 . ] in fig . [ eta - r - gamma - y ] , we plot the shear viscosity @xmath91 as a function of dimensionless shear rate @xmath191 for various wall separations @xmath39 ranging from @xmath174 to @xmath194 . in each system , @xmath91 remains constant until the applied shear rate reaches a critical value @xmath195 . the shear viscosity then decays rapidly as @xmath157 further increases , showing a typical `` shear - thinning '' behavior . [ eta - r - gamma - y ] also shows the average extension of dumbbells @xmath196 as a function of the shear rate . two comments are required here . firstly , in our mpc model , an entanglement between dumbbells is not taken into account , so that they can freely cross each other . also , the absence of an excluded - volume interaction implies that there is no benefit of a para - nematic ordering in terms of an increased sliding of parallel dumbbells along each other as in solutions of rod - like colloids ; instead , parallel dumbbells interact very similarly to isotropically oriented dumbbells , since in both cases the monomers colloide with other monomers in exactly the same fashion . thus , our system is very similar to a solution of non - interacting harmonic dumbbells , for which in the absence of a finite extensibility neither shear - thinning " nor shear - thickening " is expected@xcite , compare eq . ( [ visc_theo0 ] ) . secondly , the size of the simulation box should have no influence on the bulk viscosity at a given shear rate . however , the plateau value of the viscosity increases strongly with the wall separation @xmath39 . this indicates that boundary effects could be responsible for the observed `` shear - thinning '' behavior . ( color online ) velocity profiles for wall separations @xmath197 , @xmath198 , @xmath199 , and @xmath194 . data are obtained for spring constant @xmath76 , collision time @xmath77 , and shear rate @xmath200 . the dashed line corresponds to the applied shear rate . solid lines represent fits to the bulk part of the velocity profiles , their slopes yield the effective shear rates @xmath201 . ] we therefore examine the velocity profiles for systems with various wall separations @xmath39 . in fig . [ vp - y ] , the average velocities @xmath202 of the monomers along the flow direction are plotted as function of @xmath39 for a fixed shear rate of @xmath203 . the velocities at the boundaries deviate only very little from the wall velocities , _ i.e. _ there is very little slip at the walls , as expected . however , the velocity decays rapidly in a boundary layer of thickness @xmath204 , then decays linearly to zero at the middle plane . obviously the applied shear rate @xmath157 is not appropriate to calculate the shear viscosities from the stress tensor @xmath205 by @xmath206 . an effective shear rate @xmath201 is therefore introduced instead , which characterizes the linear bulk part of the velocity profile . at a given shear rate , the larger the wall separation , the less the effective shear rate deviates from @xmath157 , since the finite - size effect is much stronger in smaller systems . the ratio @xmath207 between the applied and the effective shear rates is plotted in fig . [ gamma - y ] , as a function of @xmath208 . at lower shear rates , _ i.e. _ @xmath209 , where @xmath210 is the critical shear rate , the ratio @xmath207 is independent of the shear rate . when the applied shear rate becomes larger than this critical value , the effective shear rate @xmath201 increases more slowly than @xmath157 . ( color online ) ratios between the applied shear rates @xmath157 and the effective shear rates @xmath201 as a function of @xmath208 for the same systems as in fig . [ eta - r - gamma - y ] . ] ( color online ) master curve of the viscosity @xmath211 as a function of the effective shear rate @xmath212 on a semi - logarithmic scale . symbols are the same as in fig . [ eta - r - gamma - y ] and fig . [ gamma - y ] . for strong shear flow , _ i.e. _ @xmath213 , the viscosity increases ( `` shear thickening '' ) . the dashed line is fitted to the data with wall separations @xmath214 . in the inset , @xmath211 is plotted as a function of @xmath215 for three - dimensional systems . a @xmath216 simulation box is chosen , while the spring constant @xmath63 , the collision time @xmath13 and the average number density @xmath47 are the same as in the two - dimensional systems . ] consequently , the effective shear viscosity can be calculated by @xmath217 in fig . [ eta - eff - y ] , @xmath218 is shown against @xmath219 for various wall separations @xmath39 . the data for different system sizes now all fall onto a single master curve , which describes the bulk shear viscosity . now , instead of `` shear thinning '' shown in fig . [ eta - r - gamma - y ] , a very weak `` shear thickening '' behavior is observed when @xmath220 . three - dimensional simulations are also carried out for systems of @xmath216 boxes along the @xmath37 , @xmath36 , and @xmath43 directions . for the same parameters @xmath76 and @xmath77 , weak `` shear thickening '' behavior is also observed , as shown in the inset of fig . [ eta - eff - y ] , when @xmath215 reaches the critical value , @xmath221 . the value of the critical shear rates are found to be very similar in two and three dimensions . [ eta - eff - y ] shows that the effective shear viscosity @xmath211 is nearly independent of the shear rate for @xmath222 . this critical shear rate corresponds to the onset of the apparent `` shear thinning '' observed in fig . [ eta - r - gamma - y ] , as well as the deviation of @xmath223 from its low - shear - rate value in fig . [ gamma - y ] . it should be noticed that the value of @xmath221 implies @xmath224 is in the range @xmath225 $ ] for system sizes @xmath226 $ ] , compare fig . [ gamma - y ] . however , it is important to note that there is already a pronounced alignment and stretching of the dumbbells for smaller shear rates ; fig . [ angle ] shows that the inclination angle @xmath188 has decreased from @xmath189 in the absence of shear flow to @xmath227 at @xmath228 , while fig . [ eta - r - gamma - y ] indicates that @xmath229 at @xmath228 . the spring constant @xmath3 of the dumbbells is of great importance , since it controls the elasticity of the fluid . we have therefore examined velocity profiles of systems of dumbbells with various spring constants . in fig . [ vp - k ] , the simulation results are plotted for a fixed applied shear rate @xmath230 . the effect of the boundary layer becomes more pronounced with decreasing spring constant . by fitting the linear parts of the velocity profiles , we find that , for the same shear rate @xmath230 , the effective shear rate @xmath212 for dumbbells with @xmath231 is about 10 times lower than that with the highest spring constant studied here , @xmath232 . the thickness of the boundary layer is proportional to the equilibrium average extension @xmath233 , as shown in the inset of fig . [ vp - k ] . ( color online ) velocity profiles for various spring constants , ranging from @xmath231 to @xmath232 . the wall separation in each case is @xmath197 . the dashed line corresponds to the applied shear rate @xmath230 , while solid lines are the fitted effective velocity profiles . the inset shows the thickness of the boundary layer @xmath204 as a function of equilibrium extension @xmath234 . the dashed line is a linear fit . ] ( color online ) the zero - shear viscosity @xmath211 as a function of spring constant @xmath63 in both two - dimensional ( circles ) and three - dimensional ( squares ) systems . the solid lines are linear fits , the dashed lines indicate the theoretical predictions ( [ eq : visc_theo_mpc ] ) . in all simulations , the collision time is @xmath77 . two - dimensional simulations are performed in systems of @xmath235 boxes , while three - dimensional simulations are in @xmath236 boxes along the @xmath37 , @xmath36 and @xmath43 directions , respectively . the average number density in three - dimensional systems is @xmath237 , which is half of value in two - dimensional systems . ] the zero - shear viscosity @xmath211 is found to depend linearly on @xmath238 , see fig . [ 3d - k ] . as @xmath3 increases , the effective viscosity @xmath211 approaches the expected value of system of point particles with mass of @xmath152 and density @xmath239 . the same linear relationship between @xmath211 and @xmath238 is also obtained in three - dimensional systems , as shown in fig . [ 3d - k ] . not only the linear dependence of @xmath211 on @xmath238 but also the prefactors are in very good agreement with the theoretical predictions ( [ eq : visc_theo_mpc ] ) . the scaled mean free path @xmath161 , which determines how far a point particle travels between collisions , is another important parameter which affects the shear viscosity . we always employ small mean free paths@xcite , so that the collisional viscosity is dominant compared to the kinetic viscosity . the data of fig . [ dt - y](a ) demonstrate that the zero - shear viscosity increases linearly with @xmath240 , for all spring constants @xmath63 studied here , as it does for a system of point particles@xcite . however , the slope decreases with increasing @xmath63 , in good agreement with the analytical results obtained from eq . ( [ eq : visc_theo_mpc ] ) , as shown in the inset of fig . [ dt - y](a ) . ( color online ) the effective shear viscosity @xmath211 as a function of the dimensionless effective shear rate @xmath241 , on a double - logarithmic scale . ( a ) for fixed spring constant @xmath242 and various mean free paths @xmath243 , @xmath244 , @xmath245 , @xmath246 and @xmath247 . ( b ) for fixed mean free path @xmath248 and various spring constants @xmath242 , @xmath249 , @xmath250 and @xmath251 . in both cases , the wall separation is @xmath197 . the inset in ( a ) shows the zero - shear viscosity @xmath211 as a function of @xmath240 . the solid line indicates the theoretical result for @xmath252 , while the other lines show the predictions ( [ eq : visc_theo_mpc ] ) for @xmath242 , @xmath249 , and @xmath251 . ] the weak `` shear - thickening '' behavior is observed for all mean free paths investigated here , see fig . [ dt - y ] . thus , this weak `` shear - thickening '' behavior is intrinsic to the mpc algorithm , and can not be avoided by a variation of the collision time . [ dt - y](a ) indicates that the critical shear rate @xmath253 depends only very weakly on the mean free path @xmath161 . therefore , we present the simulation data in fig . [ dt - y ] as a function of @xmath254 , since @xmath255 is independent of @xmath161 , while @xmath256 decreases linearly with @xmath161 . the `` shear - thickening '' behavior becomes more pronounced and slowly shifts to smaller values of @xmath254 for system of dumbbells with smaller spring constants , see fig . [ dt - y](b ) . in the range of investigated spring constants and mean free paths , the shear thickening occurs roughly at @xmath257 . it is important to note that the viscosity of the standard point - particle mpc fluid is also not independent of the shear rate , but shows a weak _ shear - thinning _ behavior at high shear rates @xcite . for our model parameters and in two dimensions , this shear - thinning behavior sets in at a shear rate @xmath258 . thus , with increasing @xmath63 , shear thickening occurs at a slowly increasing @xmath253 for @xmath259 ; for larger spring constants @xmath260 , shear thinning is observed instead , and @xmath253 decreases again ( since @xmath261 and @xmath262 for @xmath263 ) . another way to explore the viscoelastic properties of a fluid is to apply a small - amplitude oscillatory shear flow . we use here the strain amplitudes @xmath264 in the range @xmath246 to @xmath249 to mimic a small amplitude shearing . the frequencies @xmath265 ranges from @xmath266 to @xmath267 in our simulations , which provides a wide range of shear rate from @xmath268 to @xmath269 . the storage and loss moduli as a function of oscillation frequency are plotted in fig . [ y - g ] . similarly to the simulations of steady shear flow , effective shear rates are measured from the bulk velocity profiles at times when @xmath270 . by doing so , all the simulation data fall onto master curves at various wall separations from @xmath197 to @xmath194 . as can been seen from fig . [ y - g](a ) , the storage modulus @xmath121 is well fitted by eq . ( [ mg1 ] ) , indicating that the dumbbell system exhibits a typical behavior of a maxwell fluid . the relaxation frequency @xmath124 obtained from the fit of the storage modulus @xmath121 against @xmath115 is then used in eq . ( [ mg2 ] ) to fit the loss modulus @xmath122 . in fig . [ y - g](b ) , at low frequencies , @xmath271 , the simulation data follow the expected linear @xmath115-dependence very well . in this linear regime , the shear viscosity is then calculated by @xmath272 , which yields @xmath273 , in excellent agreement with the result in steady shear flow , compare fig . [ eta - eff - y ] . note that the fitted values for the amplitude @xmath125 in eqs . ( [ mg1 ] ) and ( [ mg2 ] ) differ by about a factor 2 . this indicates that the system investigated here does not behave exactly like a simple maxwell fluid . ( color online ) ( a ) storage @xmath121 and ( b ) loss modulus @xmath122 , as a function of oscillation frequency @xmath115 on a double - logarithmic scale , for systems with various wall separations ranging from @xmath197 to @xmath194 . the spring constant and the collision time are @xmath76 and @xmath77 , respectively . the dashed line in ( a ) is fitted by the maxwell model , eq . ( [ mg1 ] ) , on the basis of all simulation data , while the one shown in ( b ) is based on data for oscillation frequencies @xmath274 . ] ( color online ) ( a ) the storage @xmath121 and ( b ) the loss moduli @xmath122 , as function of oscillation frequency @xmath115 on a double - logarithmic scale , for systems of dumbbells with various spring constants ranging from @xmath76 to @xmath275 . the wall separation and the collision time are @xmath197 and @xmath77 , respectively . the inset shows the fitted relaxation frequencies @xmath124 as a function of the frequency @xmath256 predicted by eq . ( [ eq : omega_h ] ) . the dashed line shows the identity @xmath276 . ] in fig . [ k - g ] , we examine the storage and loss moduli of system of dumbbells with various spring constants . as in fig . [ y - g ] , simulation results are all well fitted by the maxwell equations ( [ mg1 ] ) and ( [ mg2 ] ) , except for somewhat different amplitudes @xmath125 . the relaxation frequency @xmath124 is found to agree very well with @xmath256 , as shown in the inset of fig . [ k - g ] . at lower oscillation frequency in fig . [ k - g](b ) , the viscosities calculated from @xmath277 are @xmath278 , @xmath279 and @xmath280 for systems with @xmath242 , @xmath249 and @xmath250 , respectively . these values are again in excellent agreement with those calculated from eq . ( [ etaeff ] ) in steady shear flow . for all spring constants @xmath63 , the fitted amplitudes @xmath125 for the storage moduli @xmath121 are about half of those calculated for the loss moduli @xmath122 . this indicates that even for a system of dumbbell with high spring constant , a simple maxwell model is not appropriate for a quantitative description . the viscosity of a simple mpc - at@xmath64 fluid ( with angular - momentum conservation ) is about a factor @xmath281 smaller than of a mpc - at@xmath59 fluid @xcite . we thus expect the viscosity of the dumbbell fluid to be affected by angular - momentum conservation as well . the simulation results for both mpc - at@xmath59 and mpc - at@xmath64 methods are compared in fig . [ fig : eta_rho ] . we find that the effective zero - shear viscosity @xmath282 increases linearly with the monomer density @xmath47 for @xmath283 . the corresponding theoretical results ( [ eq : visc_theo_mpc ] ) are in good agreement with the simulation results for both investigated spring constants . minor deviations from the linear relationship of @xmath282 with @xmath47 originate from the variation of the diffusion constant at low densities , which approaches a constant value for high @xmath47 . the viscosity of mpc - at@xmath64 is lower than for mpc - at@xmath59 , although this effect is less pronounced than for pure point - particle mpc fluids , since the main contribution to the viscosity originates from the spring tension . ( color online ) effective shear viscosities @xmath282 as a function of the density @xmath47 for mpc - at@xmath59 and mpc - at@xmath64 , each for spring constants @xmath242 and @xmath284 . the lines represent the theoretical results obtained from eq . ( [ eq : visc_theo_mpc ] ) . the wall separation and the collision time are @xmath285 and @xmath286 , respectively . ] ( color online ) scaled average of the dumbbell extension , @xmath287 , as a function of the effective shear rate @xmath288 for angular - momentum conserving and non - conserving methods . the spring constant and collision time are @xmath242 and time step @xmath286 , respectively , the density is @xmath237 and the wall separation is @xmath285 . the dashed line represents the theoretical result ( [ eq : r2 ] ) . ] in fig . [ fig : r2 ] , we present the average squared dumbbell extension , determined in the bulk as a function of the effective shear rate , along with the theoretical results ( [ eq : r2 ] ) , for both the mpc - at@xmath59 and mpc - at@xmath64 methods . note that the diffusion constant @xmath141 in eq . ( [ eq : r2 ] ) is different for angular - momentum conserving and non - conserving methods . the angular - momentum conservation has only little effect on the spring extension ; for a given effective shear rate , the extension is slightly lower for the angular - momentum conserving method . the agreement of the simulation data with the theoretical result ( [ eq : r2 ] ) is again remarkably good . as a further test for the correct calculation of the effective viscosity by the procedure described in secs . [ sec : stress ] and [ sec : steadyflow ] , we have also determined the viscosity from poiseuille flow . as in ref . , we apply a gravitational force of strength @xmath110 parallel to the walls , with @xmath110 in the range from @xmath289 to @xmath290 ( in units of @xmath291 ) . we fit the central part of the velocity profile to a parabolic flow curve . the value of this curve at the wall positions determines the effective wall slip . when this slip velocity is subtracted , eq . ( [ poiseuille ] ) in sec . [ sec : stress ] is employed to determine the viscosity@xcite . we have used this method for a system of dumbbells with @xmath242 in a @xmath292 box . excellent agreement between the two methods to calculate the zero - shear viscosity is obtained . our results for the dependence of the inclination angle @xmath188 on the shear rate can be compared with the decay of the inclination angle of flexible and semi - flexible polymers . for dilute polymer solutions in the asymptotic regime of high shear rates ( where the finite extensibility is important ) , @xmath188 has been predicted from brownian dynamics simulations @xcite and theory @xcite to decay with a power law @xmath293 and @xmath294 , respectively . for extensible dumbbells , the theory of ref . predicts@xcite @xmath295 , in excellent agreement with our simulation results . the wall slip in polymer melts has been studied extensively . in this case , molecular dynamics simulations of polymer fluids with lennard - jones interactions between monomers give a wall slip with a boundary layer thickness , which is on the order of the monomer diameter @xmath296 or less @xcite . our model could be compared more easily with results for polymer solutions , because our model does not include excluded - volume interactions . however , there is little knowledge about semi - dilute polymer solutions near a wall under flow conditions . nevertheless , some comparisons with polymer melts with moderate chain lengths are possible , where entanglement effects are absent . for example , the molecular dynamics simulations of zhang et al.@xcite show a maximum of mean squared radius of gyration at a finite distance @xmath297 from the wall , which shifts from @xmath298 for chains with @xmath299 monomers to @xmath300 for @xmath174 monomers . a multi - particle collision dynamics ( mpc ) algorithm has been developed to investigate the viscoelastic properties of harmonic - dumbbell fluid in shear flow . the method is based on alternating streaming and collision steps , just as the original mpc method for newtonian fluids . the only modification is to replace the ballistic motion of fluid point particles by harmonic oscillations during the streaming step . in this model , the entanglement between dumbbells is neglected . moreover , the storage and loss moduli are calculated by introducing a small amplitude oscillatory shear flow . our results can be summarized as follows : first , under steady shear flow , the dumbbells keep their isotropic distribution at low shear rates , but get highly stretched and orientated along the flow direction at high shear rates . the velocity profile is not uniform along the gradient direction , but boundary layers of high shear develop near the walls . the thickness of these boundary layers is found to scale with the size of the dumbbells in the absence of flow . the effective shear viscosity , calculated from the ratio between the off - diagonal component of the stress tenor , @xmath205 , and the effective shear rate @xmath201 , expresses a very weak `` shear thickening '' behavior at high shear rates . second , the dependence of the viscosity on two parameters , the spring constant @xmath63 of the dumbbells and the collision time @xmath13 , has been investigated . these two parameters are of central importance , since the former controls the elastic energy of the system , while the latter determines the mean free path @xmath161 , which measures the fraction a the cell size that a fluid particle travels on average between collisions . we find that the shear viscosity of the dumbbell fluid increases linearly with @xmath238 and with @xmath13 . third , the storage and loss moduli of our viscoelastic solvent are studied by imposing an oscillatory velocity on the two solid walls . the storage modulus @xmath121 is found to be proportional to @xmath301 at low frequencies , and to level off at @xmath276 . its behavior over the whole frequency range studied here is well described by a maxwell fluid . the loss modulus @xmath122 increases linearly with @xmath115 for low frequencies . the shear viscosities obtained from the ratio @xmath302 at low shear rates agree very well with those obtained from simulations with steady shear . on the other hand , for @xmath303 , we find that the data approach a plateau value , while for a maxwell fluid @xmath122 would decrease again for higher frequencies . our numerical results are quantitatively in good agreement with a simple theory , based on the kinetic theory of dilute solutions of dumbbells , where the transport coefficients of the standard mpc point - particle fluid are employed for the viscosity and the diffusion constant of the solvent . in our mpc algorithm of harmonic dumbbells , both elastic and viscous behaviors of solvent particles can be modeled properly , while hydrodynamic interactions are efficiently taken into account . these are valuable assets to guide future simulations on investigating rheological properties of suspensions of spherical , rod - like or polymeric solute molecules in viscoelastic fluids . we thank r.g . winkler , m. ripoll and h. noguchi for many stimulating and helpful discussions . partial support of this work by the dfg through the sonderforschungsbereich tr6 `` physics of colloidal dispersion in external fields '' is gratefully acknowledged . combining equations ( [ wall1 ] ) and ( [ wall2 ] ) , the density profile , when attractive wall potentials are introduced , can be solved analytically . considering the symmetry of the density profile , @xmath304 , only the initial half part need to be taken into account . for @xmath305 , we then arrive at which implies @xmath307 \left[\mathrm{erf}\left ( \frac{-c_2{k_{\mathrm{b}}t}+ c_1\mathrm{r}_0^{(1)}y { \mathrm{k}}/{k_{\mathrm{b}}t } } { c_1\mathrm{r}_0^{(1 ) } \sqrt{2{\mathrm{k}}/ { k_{\mathrm{b}}t } } } \right ) \right . \right . \nonumber \\ & + & \left . \left . \mathrm{erf}\left(\frac { c_2{k_{\mathrm{b}}t}+ 2(c_1\mathrm{r}_0^{(1)}-y ) c_1\mathrm{r}_0^{(1 ) } { \mathrm{k}}/ { k_{\mathrm{b}}t } } { c_1\mathrm{r}_0^{(1 ) } \sqrt{2{\mathrm{k}}/ { k_{\mathrm{b}}t } } } \right ) \right ] \right\ } \ ; , \nonumber \end{aligned}\ ] ]	in order to investigate the rheological properties of viscoelastic fluids by mesoscopic hydrodynamics methods , we develop a multi - particle collision dynamics ( mpc ) model for a fluid of harmonic dumbbells . the algorithm consists of alternating streaming and collision steps . the advantage of the harmonic interactions is that the integration of the equations of motion in the streaming step can be performed analytically . therefore , the algorithm is computationally as efficient as the original mpc algorithm for newtonian fluids . the collision step is the same as in the original mpc method . all particles are confined between two solid walls moving oppositely , so that both steady and oscillatory shear flows can be investigated . attractive wall potentials are applied to obtain a nearly uniform density everywhere in the simulation box . we find that both in steady and oscillatory shear flow , a boundary layer develops near the wall , with a higher velocity gradient than in the bulk . the thickness of this layer is proportional to the average dumbbell size . we determine the zero - shear viscosities as a function of the spring constant of the dumbbells and the mean free path . for very high shear rates , a very weak `` shear thickening '' behavior is observed . moreover , storage and loss moduli are calculated in oscillatory shear , which show that the viscoelastic properties at low and moderate frequencies are consistent with a maxwell fluid behavior . we compare our results with a kinetic theory of dumbbells in solution , and generally find good agreement .
coupled harmonic chains with short and long range interactions are ubiquitous in science and engineering . their application to calculate the phononic heat capacity by einstein @xcite marks the birth of solid state physics . beyond physics , harmonic chains feature in chemistry and biology , where they are used to model behaviour of macro - molecules , such as dna @xcite and cell membranes @xcite . in the last decade harmonic systems have been revised using techniques developed in quantum information science to study correlation properties in the quantum regime and particularly at small temperatures @xcite . thermodynamics has been very successful in characterising `` standard '' phase transitions that occur at finite temperature when a macroscopic parameter , such as pressure , is changed @xcite . quantum phase transitions ( qpts ) appear at zero temperature @xcite and are due to the change of an external parameter , such as the trapping potential of an ion trap . these transitions are driven by quantum fluctuations and have been linked to entanglement for the case of finite dimensional systems @xcite . in this paper we study a qpt in a continuous variable system : a system of trapped ions which we model as a harmonic lattice @xcite . the ions interact via a long - range coulomb repulsion and are trapped by two external potentials , see fig . [ fig : trap ] . they align in a linear configuration for big enough transversal trapping potential , @xmath0 however when @xmath0 is decreased the system undergoes a phase transition and the new equilibrium state forms a zig - zag configuration . this model is motivated by ion trap experiments @xcite , where such a qpt occurs @xcite . recent analytical studies of the transition using landau theory @xcite allowed to determine the system s classical behaviour at the transition point . moreover , the numerical treatment of the quantised system of a few ions promised the possibility of simulating linear and nonlinear klein - gordon fields on a lattice @xcite . however , a comprehensive analytical study of the quantised system has so far been lacking due to the complexity of the system . here we model the ion trap scenario as a lattice of harmonically coupled oscillators and present a first quantitative characterization of the entanglement inherent in both ground state configurations of the ions . for finite dimensional systems qpts of first ( second ) order are characterised by a discontinuity ( a discontinuity in or divergence of the first derivative ) of the negativity @xcite . we show that also in the here considered continuous variable system the structure of entanglement , measured by the negativity and the von neumann entropy , changes abruptly at the critical point and indicates the occurrence of a qpt , in a similar way as classical correlations indicate standard phase transitions . the first derivative of the negativity between two neighbouring ions has a finite discontinuity and the von neumann entropy of contiguous blocks of a single ion , two and three ions all show a divergence in the first derivative . the long - ranged nature of the coulomb interaction leads to an increase of the block entropy with increasing block size . this is in contrast to models with only nearest neighbour interaction where ` area laws ' apply @xcite and entanglement does not increase with block size , i.e. volume , as long as the surface of the block is constant . however , our results show that the increase in block entropy with the block size is quiet small due to the fast decline of the coulomb potential . the quantum fluctuations that cause the qpt are most dominant at zero temperature . however , as experiments are performed at small , but finite temperatures it is important to know how temperature affects these fluctuations @xcite . in the final part of our paper we discuss thermal states and find that the sharpness of the qpt , indicated by the entanglement , fades out with increasing temperature . another macroscopic consequence of quantum fluctuations is the lowering of the energy of the system @xcite . we compute up to which temperature the thermal state has a lower energy than any separable state . this temperature witness could be implemented in an ion trap experiment by measuring the average energy ( i.e. mean excitation ) . although the model is motivated by ion traps realisations include the vibrational motion of molecules @xcite ( nuclei in the electronic potential ) and optical lattices @xcite . we consider the hamiltonian , see fig . [ fig : trap ] , @xmath1 where @xmath2 are the coulomb potentials of the sites with @xmath3 absolute coordinates of the sites and @xmath4 are the equilibrium positions in @xmath5 direction . @xmath6 is the number of particles , @xmath7 is the charge of the ions and @xmath8 ( @xmath0 ) are the trapping potentials in @xmath5 ( @xmath9 ) direction . we assume periodic boundary conditions , @xmath10 . sketch of the harmonic lattice under consideration : each site is trapped by two external potentials , @xmath8 in @xmath5 direction and @xmath0 in @xmath9 direction . for clearness only the nearest neighbour coupling is indicated , however all ions interact via a long - range potential which we approximate harmonically . this could be for instance a coulomb potential such as is common in ion experiments . the equilibrium distances between the sites are assumed to be equidistant , with lattice constant @xmath11 in @xmath5-direction and @xmath12 along the @xmath9-direction . in @xmath13 , the transverse trapping potential @xmath0 is larger than the critical value and the ions arrange linearly . as displayed in @xmath14 , decreasing the transverse trapping potential below the critical value @xmath15 leads to a qpt causing the ions to move outwards and form a two - dimensional zig - zag structure . , scaledwidth=70.0% ] to calculate the entanglement measures we approximate the coulomb potential to second order and expand about the equilibrium positions . the key step is then to diagonalise the hamiltonian into a set of uncoupled modes , the lattice vibrations , with which analytic expressions for the measures can be obtained . similar to the classical calculation @xcite , we use a simplified model with equidistant equilibrium position in @xmath5 direction , spaced by the lattice constant @xmath11 . such condition can be realised for the central ions of a long ion chain inside a linear paul trap @xcite or for ions confined in a ring of large radius @xcite . for big trapping potential the sites are arranged on a single line , i.e. @xmath16 and @xmath17 , while for small enough @xmath0 , the equilibrium positions become @xmath16 and @xmath18 and a two - dimensional zig - zag configuration is formed . the equation determining the deviation @xmath12 in @xmath9 direction is obtained by summing the linear terms over all sites and requiring it to vanish , @xmath19 where @xmath20 numbers the neighbours of each sites . the harmonically approximated hamiltonian becomes @xmath21 with @xmath22 and @xmath23 the deviations from equilibrium . furthermore , the @xmath24 denote the second order taylor coefficients of the coulomb potential which are , for the linear and zig - zag configuration , @xmath25 we are interested in the behaviour of the entanglement between the sites in the chain for varying transverse trapping potential , @xmath0 , and particularly at the point of criticality , @xmath15 . we calculate the _ negativity _ , @xmath26 , for two modes regardless of all others , for instance , the @xmath5 degrees of freedom of two ( neighbouring ) sites . to measure the correlation of one mode with all other modes , for instance the entanglement between the @xmath9 degree of freedom of a single site or block of sites with all the other degrees of freedom in the chain , the _ von neumann entropy _ , @xmath27 , other entanglement measures are available , such as the entanglement of formation @xcite . yet they are very hard to calculate in this continuous variable scenario and we will be content with the two measures as stated . to see the effect of the long - range interaction we compare the full long - range ( lr ) coulomb hamiltonian with a cut - off version in which only nearest neighbours interact ( nn ) and the interaction with more distant neighbours is set to zero . for the _ linear configuration _ the hamiltonian decouples into @xmath5 and @xmath9 part . a discrete fourier transformation for the @xmath5 ( similar in @xmath9 direction ) of the form @xmath28 maps the space coordinates of the sites into diagonal modes , the lattice vibrations or phonons . the diagonal hamiltonian is then @xmath29 where @xmath30 , and similarly @xmath31 , are the number operators in @xmath5 and @xmath9 direction for mode @xmath32 , and the frequencies are @xmath33 where @xmath34 . the asymmetry of the system is reflected in the dispersion relations . @xmath35 is always real , whereas @xmath36 would become complex for small values of transverse trapping potential @xmath0 . this is where the quantum phase transition occurs . the critical value @xmath37 for lr interaction is @xmath38 while for nn interaction it is @xmath39 . to diagonalise the hamiltonian in the emerging _ zig - zag configuration _ , where the previously independent phonons now couple in @xmath40-direction , see eq . ( [ eq : hami ] ) , we need to amend the transformation in the @xmath9 direction eq . ( [ eq : ftlinear ] ) and transform with @xmath41 the additional factor @xmath42 compensates the alternating sign of the @xmath40 coupling in the hamiltonian . expressed with a coupling matrix @xmath43 the hamiltonian can now be written as @xmath44 with coefficients @xmath45 , @xmath46 and @xmath47 . for each @xmath32 we need the symplectic transformation @xcite , denoted by @xmath48 , that diagonalises the matrix @xmath49 , @xmath50 where @xmath51 are the symplectic eigenvalues of @xmath49 . these transformations are given by @xmath52 where @xmath53 and @xmath54 . the new normal modes are @xmath55 and using the number operators @xmath56 , and similarly for @xmath57 , we find the fully diagonalised hamiltonian for the zig - zag configuration @xmath58 $ ] . the calculation of the two - site negativity requires the evaluation of the covariance matrix of the partially transposed state of the two sites @xcite . in two dimensions this is a @xmath59 matrix of which the symplectic eigenvalues have to be found . however , in the linear configuration of the ions , @xmath5 and @xmath9 degree of freedom completely decouple and it is sufficient to consider two @xmath60 covariance matrices independently . in contrast the zig - zag configuration contains @xmath61-coupling terms and it is _ a priori _ necessary to consider the full @xmath59 matrix . as a result no entanglement occurs between the @xmath5 and @xmath9 direction in the linear configuration while the zig - zag configuration could sustain @xmath61 entanglement . however , we found that all expectation values coupling the @xmath5 and @xmath9 direction , i.e. @xmath62 , @xmath63 etc . , vanish also in the zig - zag configuration . to characterise the entanglement between two sites two sets of each two conditions , as used in @xcite , emerge . for each site @xmath64 these separability conditions are @xmath65 and similarly for the @xmath9 direction . the expectation values needed here can be calculated using the transformation rules into the diagonal modes , eq . ( [ eq : ftlinear ] ) in the linear chain , and eq . ( [ eq : trafo - zig - zag ] ) and eq . ( [ eq : diamodes ] ) in the zig - zag configuration . if one of the inequalities is violated then entanglement exists between the @xmath64-th site and its @xmath66 s neighbour and the negativity , @xmath67,\ ] ] measures their degree of entanglement . the two criteria @xmath68 and @xmath69 witness two types of entanglement . for example , the epr pair originally considered in @xcite shows violation for @xmath69 but not @xmath68 . the von neumann entropy of a single site @xmath64 in either @xmath5 or @xmath9 dimension is obtained following @xcite with the formula @xmath70 where @xmath71 , and similiarly for the @xmath9 direction , is the symplectic eigenvalue of its reduced state . to evaluate the entropy of a block of @xmath72 neighbouring sites the block entropy is then simply @xmath73 where the sum is taken over all @xmath72 symplectic eigenvalues @xmath74 in either @xmath5 or @xmath9 dimension within the block . to obtain the negativity in @xmath5 direction at zero temperature in the thermodynamical limit in the linear configuration , we evaluate eq . ( [ entcrit ] ) in the ground state leading to @xmath76 and similarly for the entanglement in @xmath9 direction . for the zig - zag configuration transformation [ eq : diamodes ] gives a more complicated expression for the @xmath77 criteria . the formula is long and not enlightening , therefore we omit it . in the thermodynamical limit we substitute @xmath78 , @xmath79 , and replace the sums with integrals , @xmath80 these integrals can be evaluated numerically and the plots are shown in fig . [ fig : all ] ( red - solid line ) . for one special set of parameters the above expression can be easily calculated analytically , namely for @xmath9 direction in the linear configuration at the critical point @xmath81 for nn interaction . the @xmath68 criterion between nearest neighbours becomes @xmath82 which leads to a value of the negativity of @xmath83 . in a similar fashion , the single - site von neumann entropy for both , @xmath5 and @xmath9 direction , can be evaluated in the thermodynamical limit . here we show the calculation for the @xmath9 direction . the symplectic eigenvalue for a single site in @xmath9 direction is @xmath84 . again we substitute sums with integrals and @xmath78 , which gives an integral expression for the symplectic eigenvalue @xmath85 at the critical point @xmath81 and for nn interaction this can be simplified to @xmath86 the first integral diverges and hence the symplectic eigenvalue and the von neumann entropy diverge at the qpt . [ fig : all ] displays both entanglement measures for decreasing transverse trapping potential @xmath0 . in the upper plots ( a and b ) both measures for the @xmath5-entanglement are constant in the linear regime . this is because the phonons in @xmath5 direction are independent of the trapping in @xmath9 direction . at the critical point both negativity and entropy are not differentiable . decreasing the trapping potential beyond the critical value , where the zig - zag configuration is formed , the @xmath5-entanglement is reduced due to the emerging @xmath40 coupling . because each site in ( b ) couples to several different neighbours , the entanglement between two distinct sites disappears . in the lower plots ( c and d ) both entanglement measures for the @xmath9-entanglement grow with decreasing trapping potential @xmath0 in the linear configuration . the even numbered ions oscillate exactly out of phase with the odd numbered ions due to the repulsive coulomb potential . the smaller @xmath0 , the larger these quantum fluctuations around equilibrium position become . at the critical point the fluctuations become strong enough for causing the ions to move outward . the entropy diverges and the negativity reaches its maximal value of @xmath83 where it is not differentiable . for the nearest neighbour coupling ( a and c ) the negativity and entropy show qualitatively the same behaviour . this can be understood easily as any entanglement of an site with the rest of the chain , measured by the von neumann entropy , is created by the coupling with only the nearest neighbours . for the long - range interaction ( b and d ) , where a single site couples to all other sites , there are significant differences between the two entanglement measures . while the negativity of two nearest neighbour sites vanishes after a threshold value of @xmath0 , each single site remains entangled with the rest of the chain , as seen by the positive value of the von neumann entropy . both entanglement measures are functions of the eigenvalues of the hamiltonian . therefore abrupt changes in entanglement can signal non - analyticity of ground state energy , which is associated with qpt s . in @xcite it was shown for finite dimensional systems that ( under certain conditions ) a discontinuity in or divergence of the first derivative of the negativity is both necessary and sufficient to signal a qpt . it seems intuitive that a similar characterisation holds also for continuous variable systems . here the divergence of entropy and finite discontinuity of the first derivative of negativity shows a qpt of second order . after the phase transition the violated negativity criterion switches from @xmath68 to @xmath69 . lowering the trapping potential further leads to another `` critical '' point @xmath87 at which the negativity becomes zero while the entropy reaches its minimal ( non - zero ) value . this graphic shows the change of trapping potential in @xmath9 direction . different geometries favour different momenta , as indicated with the arrows.,scaledwidth=50.0% ] the additional critical point @xmath87 ( and @xmath88 in ( b ) ) only appears when the interaction is harmonic as is the case in our second order approximation of the coulomb model . the point is due to a sign change of the second order coefficient @xmath89 ( @xmath90 ) at @xmath87 ( @xmath88 ) . as a consequence the interaction switches from repulsive to attractive . when anharmonic terms are taken into account , as in the numerical treatment in @xcite , these points vanish . an intuitive way of understanding the switching between @xmath68 and @xmath69 is as follows , see fig . [ fig : potentials ] . decreasing the trapping potential changes the relative strengths of the inner and outer potential for the motion in the @xmath9 direction . this leads to the change in phase in the relative momenta of neighbouring sites . in one configuration the relative potential favours momenta in the opposite directions , while the other configuration favours motion in the same direction . this is reflected in the change from @xmath68 to @xmath69 . however , although the negativity vanishes at these points , a single site is still entangled with the rest of the chain . this graphic shows the block entropy for different number of sites in @xmath5 ( a ) and @xmath9 ( b ) dimension for long range interaction in the thermodynamical limit . the entropy increases with number of sites . the inset ( c ) shows the entanglement connections between nearest neighbour and next nearest neigbour ( nnn ) sites . ] the block entropy measures how much entanglement exists between a block of sites of the lattice and the rest . for nearest neighbour interaction models there exist scaling laws showing that the amount of entanglement scales with the boundary area and not the volume of the reduced state @xcite . for a translational invariant chain with nn interaction there exists even a computable analytical result for the negativity of a bisected harmonic chain @xcite . here we investigate the block entropy for blocks up to three sites in the long - ranged coulomb lattice . due to the complexity of lr interactions few results are known so far . the inset ( c ) illustrates the increase of correlation across the block boundary with increasing block size . correlations stretch to nearest neighbours ( nn ) and next - nearest neighbour ( nnn ) , as shown in fig . [ fig : blockentropy ] while third neighbour entanglement is negligible . increasing the block size from one to two there are twice as many nnn connections across the boundary and hence the block entropy for two sites is expected to increase . this intuition is confirmed in a ) and b ) showing the block entropy in @xmath5 and @xmath9 direction , repectively . however , it can be seen that the while the entropy increases slightly with number of sites , no qualitative difference can be observed . this is due to the fact that the additional , long range entanglement is much weaker because the coulomb potential falls of quickly , with @xmath91 where @xmath66 is the distance of sites . in @xmath9 dimension ( b ) already the single site entropy is a good approximation for larger blocks . this is because transversal next nearest neighbour entanglement turns out to be very small . in @xmath5 direction ( a ) there are significant differences for growing block sites . here next nearest neighbour entanglement can not be neglected . although the number of connections between two and three sites is the same , the block three entropy is still higher . this might be due to multipartite entanglement . larger block sites are difficult to evaluate , as the symplectic eigenvalues of the covariance matrix become very complicated . one consequence of the entanglement of the sites is a lowering of the energy of the system @xcite . this can be seen by assuming that the thermal state of the system is separable , i.e. decoupled between modes . then an effective , single site hamiltonian can be obtained by removing all second order couplings between the different sites , i.e. @xmath92 etc .. the total hamiltonian becomes the sum of the single site hamiltonian @xmath93 with @xmath94 where @xmath95 then the thermal state takes the form @xmath96},\ ] ] with @xmath97 the inverse temperature . using the transformations eq . ( [ eq : ftlinear ] ) , eq . ( [ eq : trafo - zig - zag ] ) and eq . ( [ eq : diamodes ] ) , @xmath98 can be fully diagonalised . the internal energy @xmath99 for any separable state is bounded from below by zero point fluctuations , i.e. @xmath100 and any state having a smaller energy must be entangled between the individual sites . this graphic shows the negativity in @xmath9 direction of two neighbouring ions for decreasing trapping potential and increasing temperature ( in units of @xmath101=\sqrt{\frac{q^2}{m a^3}}$ ] and @xmath102=[\nu_t ] \frac{\hbar}{2 k_b}$ ] , @xmath103 ) . the red line indicates the critical temperature , obtained with the energy witness argument , below which entanglement of some form must be present in the chain . , scaledwidth=70.0% ] we now want to see how the ground state situation is modified at non - zero temperature . as the energy of the thermal state , i.e. the mean excitation of phonons , increases with temperature there exists a critical temperature , @xmath104 , for each value of the trapping potential at which the thermal state matches the energy bound . the negativity between the @xmath9 degrees of freedom of two nearest neighbours for nn hamiltonian is evaluated numerically and plotted in fig . [ fig : tempy ] where the critical temperature for full separability is also indicated as a red - line . when the trapping potential is lowered , the negativity increases until reaching its maximal value at the critical @xmath15 . further lowering leads to a decrease of the negativity until it vanishes at point @xmath87 . however , when further decreasing @xmath0 the negativity grows again , yet the two entanglement criteria @xmath77 are now switched . as expected , increasing the temperature leads to smaller values of negativity and smoothens the entanglement measures to make it differentiable at the critical point . for large @xmath0 the negativity is small , but remains finite until relatively high temperature . the sharp peak at the qpt remains almost constant for finite @xmath105 and decreases fast . thermal states within the red outlined area have a smaller energy than any separable state and their entanglement is therefore detected by the energy witness . remarkably , states at @xmath87 are entangled for temperatures up to @xmath106=0.12\frac { [ \nu_t ] \hbar}{2 k_b}$ ] , even though there is no nearest neighbour entanglement in @xmath9 direction . this is because there is still nearest neighbour entanglement in @xmath5 direction and possibly also multipartite entanglement the chain . in this paper , we revised the classical phase transition in a long range harmonic chain @xcite using a fully quantised model . two measures of entanglement display critical behaviour : the von neumann entropy of a single site and blocks of two and three sites diverge at the critical point while the negativity is not differentiable . thus also in this continuous variable system entanglement indicates a qpt , as previously shown for discrete systems @xcite . the negativity depends only on single site and nearest neighbour correlations ; the single site von neumann entropy even only depends on single site measurements . our calculation shows that even this local , single site function is able to detect a global change in configuration . this implies that instead of examining two point correlations functions one can alternatively consider the entanglement measures stated . this will be advantageous in experimental situations when the number of different measurement procedures is best kept as low as possible . we are aware that for the moment ion traps can not yet perform the required measurements of e.g. single site variance of space and momentum operator , but this is a technical , not a fundamental problem . furthermore , our results confirm that this phase transition is of second order as indicated in @xcite . at finite temperature , the negativity still displays critical behaviour , as seen in fig . [ fig : tempy ] , however the non - differentiable cusp fades out quickly with increasing thermal noise . tuning across a qpt provides a means of changing the amount and structure of continuous variable . experiments with ion - traps are ideally suited to study qpts with great precision . the authors wish to thank dieter jaksch for inspiring discussions . is supported by the national research foundation and ministry of education , in singapore . j.a . is grateful for support by the epsrc s qipirc program in the uk and thanks the centre for quantum technologies in singapore for hospitable stays . v.v . acknowledges financial support from the engineering and physical sciences research council , the royal society and the wolfson trust in uk as well as the national research foundation and ministry of education , in singapore .	the entanglement properties of the phase transition in a two dimensional harmonic lattice , similar to the one observed in recent ion trap experiments , are discussed both , for finite number of particles and thermodynamical limit . we show that for the ground state at the critical value of the trapping potential two entanglement measures , the negativity between two neighbouring sites and the block entropy for blocks of size 1 , 2 and 3 , change abruptly . entanglement thus indicates quantum phase transitions in general ; not only in the finite dimensional case considered in @xcite . finally , we consider the thermal state and compare its exact entanglement with a temperature entanglement witness introduced in @xcite . keywords : _ harmonic lattice , quantum phase transition , entanglement _
understanding the content and distribution of cold gas in galaxies is an important step to understand the formation and evolution of galaxies . ugc8802 is an interesting target selected from the _ galex arecibo sdss survey ( gass ) @xcite . one interesting aspect is its cold gas content . ugc8802 is a disk galaxy with stellar mass @xmath0 , while its hi mass is estimated to be as high as @xmath1 @xcite and its molecular gas mass is only about one - tenth of its hi mass @xcite . compared to other disk galaxies in this stellar mass range , the high neutral gas content of ugc8802 is uncommon @xcite . why ugc8802 has such high neutral gas fraction and how ugc8802 acquires its cold gas are still open questions @xcite . _ another interesting aspect of ugc8802 is its flat profile of the star formation rate ( sfr ) @xcite . the sfr surface densities of ` normal ' disk galaxies typically decrease ( sometimes exponentially ) from the center to the outer regions of the disk , such as the milky way , m31 , etc @xcite . besides the flat sfr profile , @xcite found that ugc8802 shows a steeper radial oxygen gradient than the average gradient of the gass sample . therefore , it is interesting to explain these observed results under the frame of the current models of galaxy formation and evolution . for individual disk galaxy , parameterized modeling is proven to be fruitful in exploring their galactic formation and evolution @xcite . in this letter , we construct such a parameterized model to investigate the radial - dependent star formation history ( sfh ) of ugc8802 and so as to explore its growth history . our model is based on that of @xcite . here we emphasize main ingredients of the model , especially some revisions . we adopt the standard cold dark matter ( cdm ) cosmology with @xmath2 and @xmath3 . we assume that the galactic disk is sheet - like and composed by a set of independent rings , each 500pc wide . the disk originates through continuous gas - infall from the dark halo . we assume that the disk begins to form at 12.5 @xmath4 ago , which roughly corresponds to @xmath5 under the standard cosmology . at given radius @xmath6 , the gas infall rate @xmath7 ( in units of @xmath8 ) is assumed to be gaussian in time @xcite : @xmath9 ^ 2/2\sigma^2}\ ] ] where @xmath10 is the infall - peak time and @xmath11 is the scatter . we adopt @xmath10 to be free and set @xmath12 since the model results are insensitive to varying @xmath11 in the @xmath13 range @xcite . the @xmath14 are a set of separate quantities normalized by the stellar mass surface density of present - day @xmath15 , where @xmath16 is the cosmic age and set to be @xmath17 according to the adopted cosmology . the stellar mass of ugc8802 and its disk scale - length at the present time are adopted to be @xmath18 and @xmath19 @xcite . we assume that the galaxy ugc8802 at present is a pure - disk system and @xmath15 follows an exponential profile , and so that the central stellar mass surface density is given as @xmath20 . the star formation ( sf ) law is one of the key ingredients of our disk galaxy formation model . @xcite found a power - law relationship between the galaxy - averaged sfr surface density and the galaxy - averaged total gas surface density , which is often called as the classical kennicutt - schmidt sf law and has been widely used in the chemical evolution models of the disk galaxy formation . later , extended analyses to kpc - scale regions within several local group galaxies show that the sfr surface density correlates stronger with the surface density of molecular hydrogen than that of atomic hydrogen @xcite . in this study , we adopt a recently observed molecular surface density related sf law ( hereafter , it is termed as the @xmath21-based sf law ) , that is , the sfr surface density @xmath22 is linearly proportional to the molecular hydrogen surface density @xmath23 : @xmath24 where @xmath25 is the molecular gas depletion time . @xcite and @xcite derived a constant molecular gas depletion time as @xmath26 in their sample , while @xcite found the non - universality of @xmath25 in galaxies in the co legacy database for the gass ( cold gass ) sample . @xcite found that the strongest dependence of @xmath25 is on the stellar mass and the mean @xmath25 is parameterized as log@xmath27 . according to this relation , we adopt @xmath28 for ugc8802 . regarding the molecular - to - atomic mass fraction @xmath29 , @xcite obtained an empirical fit relation : @xmath30 where @xmath31 is the mid - plane pressure , @xmath32 and @xmath33 are fitting constants and we adopt @xmath34 and @xmath35 @xcite . the pressure is estimated to include gas self - gravity @xcite : @xmath36 where @xmath37 is the total gas surface density , @xmath38 and @xmath39 are the vertical velocity dispersion of gas and stars . we adopt @xmath40 @xcite , while @xmath39 is estimated as [ see appendix b of @xcite for details ] : @xmath41 a similar @xmath21-based sf law has been successfully adopted in the semi - analytical model of galaxy formation by @xcite . @xcite also use the @xmath21-based sf law to explore the chemical evolution and sfh of m33 and find that , comparing to the model adopting the classical kennicutt - schmidt sf law , the model adopting the @xmath21-based sf law predicts steeper color and matellicity gradients , which are more consistent with the observations . other components of the model , such as the initial mass faction ( imf ) , chemical evolution , etc . , are the same as that of @xcite . we also consider the contribution of gas outflow process by assuming that the gas outflow rate is proportional to the sfr and the coefficient is set to be @xmath42 according to the mass - dependent model of @xcite . we point out that our final results are insensitive to the variation of @xmath43 since ugc8802 is a massive galaxy and the gas outflow process is not a dominant process of its chemical evolution . we emphasize that our model only has one free parameter , the infall - peak time @xmath10 . in @xcite , @xmath44 is assumed to be a function of galactic stellar mass and do not vary with radius . in the case of ugc8802 , we assume that @xmath10 is a function of radius since we will further consider the radial profiles of the ugc8802 rather than the global properties of galaxies as that in @xcite . since the formalizing of @xmath45 regulates the shape of sfhs along the disk , we aim to get the constraints on @xmath45 from the observed properties of ugc8802 and then get insights on its growth history . we present our results step by step . firstly , we summarize main observational properties of ugc8802 , and then explore the influence of the free parameter on our results . finally , we present a viable model to discuss the radial - dependent sfh of ugc8802 . the observed data of the radial profiles of ugc8802 are adopted mainly from @xcite and plotted in fig . 1 . since our model only has the galactic - center distance and can not distinguish the negative position from the positive one , we plot the observed data of the negative and positive positions as open triangles and filled squares in fig . 1 , respectively . the nuv - r color profile of ugc8802 is measured using sdss and galex photometry @xcite , but the data has not been corrected for dust - attenuation . the observed data of oxygen abundance @xmath46 , the sfr surface density and @xmath47 are obtained from the long - slit spectroscopy of ugc8802 @xcite . the data @xmath48 are estimated from the observed sfrs and the stellar surface density @xmath15 by assuming @xmath49 fraction of the total formed stellar mass are locked in the stellar mass in the present day , where @xmath50 is the return fraction and we set @xmath51 according to the adopted imf . the comparison between model predictions of the radial profiles and the observations are presented in fig.1 . the dash and dash - dot lines are model results of two limiting cases of @xmath52 and @xmath53 , respectively . the case @xmath52 ( dash lines ) corresponds to a time - decreasing gas - infall process that most of the gas has been accreted in the early stage of its history , while that of @xmath53 ( dash - dot lines ) represents a time - increasing gas accretion that there is still a large fraction of cold gas - infall at the present time . 1 shows that the model predictions are very sensitive to the adopted @xmath10 . the model adopting the later infall - peak time ( dash - dot lines ) predicts the higher surface density of @xmath54 and @xmath55 , the higher sfr and specific sfr , the lower gas - phase metallicity , the older mean stellar age and the bluer color . these results are straightforward , the later @xmath10 corresponds to a more recent gas infall and then results in a younger stellar population and a higher fraction of cold gas in present . and @xmath53 , respectively . the solid lines plot the predictions of the viable model , which adopts @xmath56 . the observed data are described in details in the text . , title="fig:",width=302,height=302 ] + fig . 1 also shows that the area between the dash and dash - dot lines almost covers the locations of the observed data . there is a trend that the observed data at the inner region approaches the dash lines ( the early @xmath10 ) and the data at the outer regions approaches the dash - dot lines ( the late @xmath10 ) . after a set of calculations and comparisons , we select a viable model by adopting an inside - out formation scenario @xmath57 and plot its model results as solid lines in fig . 1 . this model predicts @xmath58 and @xmath59 , which in good agreement with the observed values [ @xmath60 , @xmath61 @xcite ] , considering the observed uncertainties . our model - predicted @xmath62 colors are systematically bluer than the observed ones , which is also reasonable since the observed @xmath62 color has not been corrected for dust - attenuation and our model predictions are dust free . , while the dash line shows the observed trend of size evolution of massive galaxies . the bottom panel shows the normalized evolution history of different baryonic components , including the stellar mass ( the solid line ) , the atomic hydrogen mass @xmath63 ( the dash line ) and the molecular hydrogen mass @xmath64 ( the dot line).,title="fig:",width=302,height=264 ] + the good consistence between our model predictions and the observations implies that the continue gas - infall model is also viable for the ` peculiar ' ugc8802 , which has steeper metallicity gradient , flatter sfr profile and extraordinary higher neutral gas fraction than normal disk galaxies . the key ingredient of this successful model is that the infall - peak time @xmath45 increases from @xmath65 in the innermost disk to 12.5gyr at the out region around @xmath66 . that means the formation of ugc8802 is quite late and the disk growing is still ongoing . in fact , this scenario is the well - known idea that the disk forms inside - out and has also been applied in previous models of formation and evolution of disk galaxies @xcite . the above results have shown that the inside - out growth scenario can nicely reproduce the main observed properties of ugc8802 . in fig . 2 , we show further predictions of the growth history of the ugc8802 of this established model . the upper panel of fig.2 shows the model predicted time evolution of the half - size stellar mass @xmath67 ( solid line ) , which is defined as the radius where contains half of the total stellar mass . as we can see , our model predicts that the half - size of ugc8802 increases almost linearly with time , which is also globally consistent with the observations of the average size evolution trend of the massive galaxies [ @xmath68 , @xcite , dash line ] . this size growth is actually another angle of view of the disk inside - out formation scenario . in the bottom panel , we plot the model predictions of the evolution of the different mass components , which includes the stellar mass @xmath69 ( solid line ) , the atomic hydrogen mass @xmath63 ( dash line ) and the molecular hydrogen mass @xmath64 ( dot line ) the quantities are normalized by its value at the present - day . the bottom panel of fig . 2 shows that , in the early epoch , with the continue infall of cold gas ( mainly in the form of the atomic gas ) , the molecular gas accumulates faster , while the accumulation of stellar mass is lagged behind . later , the mass of the molecular gas becomes to be asymptotic , while the growth rate of stellar mass is speeded . indeed , half of the stellar mass has been accumulated during the last @xmath70 . to sum up , our results suggest that ugc8802 is a young galaxy and still active in star formation processes . in this work , we build a bridge for disk galaxy ugc8802 between its observed properties and its growth history by constructing a simple chemical evolution model . we find that a late infall - peak time @xmath10 , which corresponds to a recent gas - infall process , is necessary to explain the observed high atomic and molecular gas content of ugc8802 . the model adopting an inside - out formation scenario @xmath57 and the @xmath21-based sf law can reproduce the observed radial profiles of ugc8802 very well . this consistency does encourage us that our simple model is viable for such a ` peculiar ' galaxy . this result implies that there should be no violent star burst process ( e.g. the major merging ) during the growth history of ugc8802 . according to our model , half of the stellar mass of ugc8802 may have been accumulated during the last @xmath70 and the sf process of the outer disk is still active . the physics behind the very young history of ugc8802 may related to its environment , however , which is beyond the scope of this letter . the success of our simple model on the case of ugc8802 implies that the observed properties of the gas - rich galaxies might also be well modeled by ` tuning the knob ' on gas inflow with a single parameter . indeed , @xcite and @xcite find that the color and metallicty of the outer region of the galaxies strongly correlate with its total atomic mass @xmath63 , i.e. , the higher the @xmath63 , the bluer the color , the lower the metallicity . in our story , the high atomic gas fraction of gas - rich galaxy may be the result of a late and smooth gas accretion history , and the bluer color and lower metallicity of its outer region are embedded in the disk inside - out formation scenario . we will further explore these questions in a forthcoming paper . we thank the anonymous referee for suggestions to greatly improve this paper . this work is supported by the national science foundation of china no.10573028 , 10403008 , the key project no.10833005 and no . 10878003 , the group innovation project no.10821302 .	the disk galaxy ugc8802 has high neutral gas content and a flat profile of star formation rate compared to other disk galaxies with similar stellar mass . it also shows a steep metallicity gradient . we construct a chemical evolution model to explore its growth history by assuming its disk grows gradually from continuous gas infall , which is shaped by a free parameter the infall - peak time . by adopting the recently observed molecular surface density related star formation law , we show that a late infall - peak time can naturally explain the observed high neutral gas content , while an inside - out disk formation scenario can fairly reproduce the steep oxygen abundance gradient . our results show that most of the observed features of ugc8802 can be well reproduced by simply ` turning the knob ' on gas inflow with one single parameter , which implies that the observed properties of gas - rich galaxies could also be modelled in a similar way .

Subsets and Splits

Data Mining Abstracts

This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.